Tools for predicting risk of mortality in the ICU setting: do we need a crystal ball or rose colored glasses?

Hematopoietic stem cell transplantation (HSCT) applied to children is associated with high risk for organ failure, ICU admission, morbidity and mortality. "Respiratory failure" after HSCT carries a historically grave prognosis. Factors associated with high risk for critical care complications in HSCT patients have been identified, but are dependent on timing and intensity of interventions. Several ICU severity of illness scoring systems predict prognosis on the basis of physiologic stability, organ system involvement, and intensity of supportive measures; but these tend to underestimate post-transplantation mortality risk. Adjustment of scoring systems and logistic regression factor analysis are promising adjuncts, but have not been adequately validated. Specific endpoints such as death, length of ICU or hospital stay, and neurologic function are relatively easy to quantify; but, quality of life is difficult to assess and report. What constitutes "heroic therapy" in one institution may qualify as "routine" care in another. Therefore, tools to predict mortality in the pediatric HSCT recipient requiring intensive care are difficult to apply to the individual patient, and remain more an art than science. This manuscript attempts to briefly define and review the pertinent types of PICU severity of illness and mortality prognosis scoring systems, and their application to pediatric HSCT patients. Pitfalls in application of physiology, organ system failure, therapeutic intensity, disease specific, and history-based scoring systems are discussed. Prospective validation studies for severity of illness systems and the evolution to concurrent registry-style data collection and analysis are necessary for the HSCT patient requiring ICU care.