Objective To determine which factors affect the mortality outcome of admissions to paediatric intensive care (PICU) in our institution.

Methods A retrospective analysis of outcome of 3629 patients admitted to PICU over a 4 year period, 2003 to 2007. A multiple logistic regression model was used and adjusted for the following factors: PIM2, PIM diagnostic group, admission time, source of admission and unplanned admission. PIM2 is the Paediatric Index of Mortality score, which estimates the risk of death and is also intended as a means of assessing outcome in PICU.

Results No factor analysed significantly predicted death within the first 24 hours apart from the PIM2 score OR 2.8 (2.1–3.7) p value<0.001. Patients admitted between 02.00 hours and 08.00 hours had significantly higher PIM2 scores and hence were more likely to die within 24 hours. After adjustment external admissions and planned admissions were less likely to die, as were patients in the Respiratory PIM diagnostic group (OR 0.57, p<0.001). Internal unplanned admissions had the highest mortality rate despite adjusting for disease severity SMR = 1.55; in this group PIM2 did not predict death within 24 hours or overall.

Conclusion PIM2 appears to overestimate mortality for respiratory patients. Outcomes for unplanned patients from within our institution are worse than predicted by PIM2; reasons are unclear but may include an inherent difference in case mix and/or differences in the performance of PIM2 as a risk model for different diagnostic groups. Pre-admission care for this group is being further evaluated to determine effect.