Prognostic scoring in critically ill children: What to predict? by ProQuest


More Info
									                            CMAJ                                                                   Commentary
                          Prognostic scoring in critically ill children: What to predict?

                          Martin C.J. Kneyber MD PhD

                          Previously published at

                          @@       See related research article by Leteurtre and colleagues, page 1181

                                  everity-of-illness scoring systems have been widely               Key points
                                  used in pediatric intensive care units (PICUs) to
                                                                                                    •    Scoring systems to predict mortality are widely used in
                                  quantify patient outcomes. These scoring systems can                   pediatric critical care.
                          be used for internal and external benchmarking to assess
                                                                                                    •    Mortality may not be the best outcome parameter to
                          quality of care, and for identifying and stratifying patients                  predict.
                          enrolled in clinical trials. Most systems use mortality as the            •    Surrogate parameters should be pursued, including the
                          outcome measure to be predicted. Alternatively, a surrogate                    functional status of the child.
                          outcome with a higher incidence than mortality may be
                          selected as an outcome parameter for scoring systems. Organ
                          dysfunction could well be such a surrogate and has been used            the PELOD score among 1806 critically ill children admitted
                          by Leteurtre and colleagues.1                                           to seven European PICUs and concluded that it was a good
                             The scoring systems most frequently used in PICUs are                predictor of mortality.7 Nevertheless, several years later, the
                          the Pediatric Risk of Mortality (PRISM) and the Paediatric              authors recognized that their external validation was erro-
                          Index of Mortality (PIM) scores. The latest version of the              neous because their model had poor calibration.8 This lack of
                          Pediatric Risk of Mortality (PRISM-III) was derived from                good calibration was recently confirmed by data from nearly
                          data from more than 11 000 patients in 32 PICUs and was                 1500 children admitted to two PICUs in Brazil and
                          recalibrated on more than 20 000 patients.2 It uses physio-             Argentina.9 In both this study and the original report of exter-
                          logic, laboratory and diagnostic data to predict mortality after        nal validation, mortality was underpredicted by the score
                          12 or 24 hours in intensive care. The latest version of the Pae-        among patients at lower risk of death and overpredicted
                          diatric Index of Mortality (PIM2) was derived from data from            among those at higher risk.
                          about 20 000 patients in 14 PICUs and uses physiologic and                 Despite this major issue, Leteurtre and colleagues have
                          laboratory data upon admission to the intensive care unit.3             reanalyzed the data that they used to validate the PELOD
                          Essentially different from the Pediatric Risk of Mortality              score in order to describe temporal changes in the score in
                          score, the Paediatric Index of Mortality originated from the            relation to mortality.1 They have attempted to quantify the
                          concept of eliminating lead-time bias (i.e., the influence that         subjective feeling of physicians that a patient is likely to die if
                          management before admission to the intensive care unit might            his or her organ function deteriorates or fails to improve.
                          have on the physiologic variables during the first 24 hours             Bearing in mind the poor calibration of the PELOD score and
                          after admission to the unit). Furthermore, it is less likely to be      several questions regarding the methodology of the study, the
                          influenced by the quality of management in the PICU during              PELOD score cannot be regarded as a suitable surrogate para-
                          the first 24 hours.                                                     meter for death. First, and perhaps most important, confound-
                             The problem is that mortality may not be the best outcome            ing by indication was introduced because variables were mea-
                          parameter, for two important reasons. First, because of the             sured only if requested by th
To top