Analysis of Severity of Diseases in Intensive Care Unit

The structure of pediatric intensive care populations of patients can differ substantially among units. The differences between the original population of patients, that the score was derived from, and populations of patients in units where the mortality prediction scoring system is applied to, as well as differences among the groups within the same patient population, can make the mortality prediction inaccurate. The parameters for the Paediatric Index of Mortality (PIM 2) score were collected prospectively for 591 consecutively patients admitted to the Pediatric Intensive Care Unit (PICU). The observed mortality (OM), the predicted mortality (PM), and the standardized mortality ratio (SMR) for diverse groups of patients were assessed separately. PM was higher than OM for surgical patients than for medical patients, and for urgent than for scheduled admissions. Within the diagnostic groups, SMR was < 1 in neurological patients, and in patients with respiratory diseases SMR was > 1. Within the age groups, SMR was > 1 in preschool children, and in adolescents SMR was < 1. When length of stay was analyzed, SMR was > 1 in long stay patients. The predictive power of the PIM 2 scoring system is not, therefore, equal for all groups of our population of patients. That means that the differences of structure of populations of patients can lead to inaccuracy of a mortality prediction model. This problem can be prominent in small units with small and uniform populations of patients.