The Pediatric Sequential Organ Failure Assessment (pSOFA) score may predict in-hospital mortality in critically ill pediatric patients
EBM Focus - Volume 12, Issue 34
- The Sepsis-3 task force recommends using the Sequential Organ Failure Assessment (SOFA) score for clinical characterization of organ dysfunction in adults, but its applicability to the pediatric population is unclear.
- A pediatric version of the SOFA (pSOFA) was developed and its performance to predict in-hospital mortality was evaluated in a retrospective prognostic cohort study including 8,711 admissions to a pediatric intensive care unit (PICU).
- The maximum of the daily pSOFA scores had strong discrimination for in-hospital mortality (c-statistic 0.94, 95% CI 0.92-0.95). Prognostic performance was statistically superior to or similar to other pediatric risk scores.
For its definition of sepsis in adults, the Sepsis-3 task force recommends that clinical characterization of organ dysfunction be conducted with the SOFA score, which uses common clinical and laboratory tests to assess neurologic, respiratory, coagulation, hepatic, cardiovascular, and renal functions (JAMA 2016). However, the applicability of the SOFA to the pediatric population is limited as it does not take into account age-dependent variations of physiologic variables. Given the SOFA’s use in the Sepsis-3 definition and its familiarity among the critical care community, a pediatric score that is analogous to and uses similar variables as the SOFA would aid in risk assessment for children and adolescents. The current study developed a pediatric version of the SOFA (pSOFA) by including age-adjusted cut-offs for cardiovascular and renal variables and adding the SpO2:FiO2 ratio as a noninvasive surrogate of lung injury. Like the SOFA, the pSOFA total score ranged from 0 to 24, with 24 indicating a high risk of poor outcomes. This retrospective prognostic cohort study evaluated the pSOFA for predicting in-hospital mortality, including comparisons to other pediatric organ dysfunction scores. The study included 6,303 critically ill patients ≤ 21 years old presenting to a multidisciplinary PICU for a total of 8,711 admissions. Electronic health records were used to calculate the pSOFA, the Pediatric Multiple Organ Dysfunction Score (P-MODS), the Pediatric Logistic Organ Dysfunction (PELOD) score, and the PELOD-2 score every 24 hours until death, discharge, or day 28 of hospitalization. Several uses of the scores, including the maximum of the daily scores and the score at admission, were evaluated.
Using definitions of sepsis and septic shock analogous to the Sepsis-3 definitions, sepsis occurred in 1,231 admissions, of whom 12% died, and septic shock occurred in 347 admissions, of whom 32% died. The maximum of the daily pSOFA scores had strong discrimination for predicting in-hospital mortality; death occurred in about 70% of patients who had a maximum score of 20-24, 50% with score 17-20, 30% with score 13-16, 8% with score 9-12, 2% with score 5-8, and 0% with score 0-4 (all values were estimated from a figure). Prognostic performance was quantified by the c-statistic, which represents the probability that a patient who died would have a higher score than a patient who did not die; a c-statistic of 1 has perfect performance and 0.5 indicates no predictive value. Using the maximum of the daily scores, the pSOFA had a c-statistic of 0.94 (95% CI 0.92-0.95), which was significantly superior to the P-MODS (0.91) and not significantly different than the PELOD (0.93) or PELOD-2 (0.94). Using the score at PICU admission, the pSOFA had a c-statistic of 0.88 (95% CI 0.86-0.91), which was statistically superior to the P-MODS (0.86), PELOD (0.86), and PELOD-2 (0.87) scores (p ≤ 0.02 for each), and not significantly different than the Pediatric Risk of Mortality (PRISM) III score using information from the first 24 hours of admission (0.88).
This study showed that the maximum of the daily pSOFA scores may have strong discrimination for predicting in-hospital mortality among patients in the PICU, with a prognostic performance statistically superior or similar to other pediatric risk scores. However, the clinical utility of the pSOFA score at PICU admission is unclear, as the in-hospital mortality rates for this usage were not reported. This first evaluation of the pSOFA used a retrospective design and data from a single PICU, limiting confidence and generalizability of these results. Additional evaluations using a prospective design with data from several centers and reporting observed mortality rates according to pSOFA scores at admission and other statistics with more direct clinical relevance will further clarify its performance and clinical utility. In the meantime, the pSOFA may be a promising pediatric risk score that is analogous to the widely used SOFA score and so may aid in assessing critically ill pediatric patients in the ICU.