Comparison of different severity score used in an oncologic pediatric intensive care unit / Comparação dos diferentes escores de gravidade utilizados em unidade de terapia intensiva pediátrica oncológica

AUTOR(ES)
DATA DE PUBLICAÇÃO

2010

RESUMO

Introduction: Although cancer is responsible for more deaths in children older than 1 year of age than any other disease, outcomes are improving. Improvements in survival rates and quality of life have been achieved due to early diagnosis, as well as aggressive therapies (chemotherapy, radiotherapy, surgery, bone marrow transplantation, and supportive measures). These advances have resulted in an increased need for critical care services for postoperative care, and complications of cancer or its therapy. Prognostic scoring tools have been used to predict outcome in patients who are admitted to PICUs. Although none of these models can be used to predict individual outcomes, they can assist physicians in discussions of prognosis and in clinical decision-making, and may improve the allocation of intensive care resources. Few studies have addressed the performance (calibration and discrimination properties) of prognostic scoring tools in cancer patients. Objective: To evaluate the performance of three mortality prediction tools (Pediatric Risk of Mortality - PRISM, Pediatric Index of Mortality - PIM, and PIM2) at a pediatric oncology intensive care unit (POICU). Method: Retrospective cohort study of all consecutive patients was admitted between January 1, 2004 and December 31, 2008; in a POICU at a tertiary care cancer hospital in Brazil. Results: All admissions to the POICU with a definite diagnosis of leukemia or tumor were included in the present study. Four hundred and fifty-one admissions among 260 patients were evaluated. The mortality rate was 5.3%. Prediction of mortality versus survival was assessed by calculating the area under the receiving operating characteristic curve (AUROC) for each model. The AUROC (95% Confidence Interval) for PRISM, PIM, and PIM2 were 0.89 (0.81-0.97), 0.81(0.73-0.90), and 0.81(0.72-0.90), respectively. The standardized mortality ratios (SMRs) were over-predicted by PRISM (SMR = 0.88) and under-predicted by PIM (SMR = 1.02) and PIM2 (SMR = 1.10). Calibration across deciles of risk intervals assessed by the HosmerLemeshow goodness-of-fit test showed c2 (8) = 9.07 (p = 0.33); c2 (8) = 10.8 (p= 0.21); c2 (8) = 12.78 (p=0.11) for PRISM, PIM, and PIM2, respectively. Conclusions: For the entire population, all three tools demonstrated adequate prediction of mortality versus survival and a good calibration assessed by the HosmerLemeshow goodness-of-fit test. PRISM score predicted mortality better than PIM or PIM2, although not significantly so

ASSUNTO(S)

mortality unidades de terapia intensiva pediátrica child neoplasms Índice de gravidade de doença prognóstico mortalidade pediatric intensive care units prognosis severity of illness index criança neoplasias

ACESSO AO ARTIGO

Documentos Relacionados