INTRODUCTION
Assessment of illness severity is crucial for a wide range of intensive care unit (ICU) uses, including quality assessment, controlling for severity of illness in clinical studies, and studies of intensive care unit resource utilization and management. There are several scores used to quantify severity of illness using probabilistic models to predict the individual mortality risk of ICU patients (e.g. APACHE II, SAPS II, CRIB, PRISM, PIM).
Although specific scores for paediatric intensive care (PIC) have been developed, such as the Pediatric Risk of Mortality (PRISM and PRIM III) (1 2) and Paediatric Index of Mortality (PIM) (3), there is currently no score available that has been validated for general use in PIC in Portugal. PRISM and PRISM III, for example, have been validated in large populations in the USA but there is conflicting evidence about their applicability elsewhere (4 5 6).
In addition to uncertainty around the accuracy of the different mortality scores, death is a rare outcome following PIC and there is concern that intensive care may be keeping individuals alive in very poor states of functional health for significant lengths of time. The implication is that morbidity is more relevant than mortality when comparing outcomes between groups of patients in PIC and that the life expectancy of the survivors may understate the impact of post PIC morbidity in children.
The current study aims to assess and optimise the performance of PRISM, PRISM III and PIM in predicting the patients' mortality risk in Portuguese PIC units and, in collaboration with a research team that is leading the same study in the United Kingdom (UK PICOS research team), to develop a model for severity of illness based on the health status 6 moths after PIC admission.

METHODS
Risk factors and survival data will be collected for all children admitted to three major PIC units during a two and a half years period (an expected total of 2000 children based in previous years). The data collected will include data for all variables necessary for the calculation of PRISM, PRISM III and PIM.
Health status will be assessed using the Health Utilities Index II (HUI II). This questionnaire will be administered to the parents of children admitted to participating PIC units, as close as possible to the time of admission as seems appropriate to the consultant in charge of the child's care. The parents will be asked to fill out the questionnaire in terms of their child's health status immediately prior to the event which lead to their admission to the PIC unit. At six months post admission to the PIC unit, the parents of all surviving children will be sent a second copy of the questionnaire and asked to complete it for their child's current health status.
A web based application will be specifically developed to register, centralize and standardize the data from the participating PIC units.

EXPECTED RESULTS
The results of this project will allow a deeper understanding of the reality of paediatric intensive care in our country and a valuable base for more accurate research in the field.