Initial correspondence from Drs. Leclerc, Duhamel, Leteurtre

Dear Editor,

We read with interest the paper by Schlapbach et al. comparing the performance of SIRS criteria with measures of organ dysfunction in children with sepsis [1]. The performance of pediatric SOFA (pSOFA) at discriminating mortality was not higher than that of the Pediatric Logistic Organ Dysfunction Score-2 (PELOD-2) [area under ROC curve (AUROC) = 0.829 vs. 0.816; p = 0.970]. Yet, and suggested by the authors themselves, as pupil size and lactatemia were not available in their database, this may have reduced the performance of PELOD-2 in their population [1]. In our recent study, AUROC of PELOD-2 including these two parameters was 0.91 in children with suspected infection and in those with low mean blood pressure and hyperlactatemia [2]. In another study in the whole population of critically ill children in one PICU, the performance of pSOFA was considered better than that of PELOD-2 (AUROC = 0.88 vs. 0.87; a difference of low relevance in our opinion) [3]. Unfortunately, it was impossible to obtain from Matics and Sanchez-Pinto additional information regarding the performance of PELOD-2 in the subgroups of children with infection and those with septic shock. Although the databases were different, AUROC of quick SOFA (qSOFA) proposed by Schlapbach et al. [1] appeared lower than that of the quick PELOD-2 (qPELOD-2) reported by us (0.74 vs. 0.82) [2]. Can Schlapbach et al. calculate the qPELOD-2 (and AUROC) on their database?

Sepsis-3 definitions in children need further validation [1, 3], and we suggest comparing newly developed pSOFA in its different versions versus validated PELOD-2 used worldwide, and qSOFA versus qPELOD-2 in future studies.

Reply from Drs. Schlapbach, Straney, Bellomo, MacLaren, Pilcher

We would like to thank Leclerc et al. for their correspondence. Recent studies attempting to adapt Sepsis-3-based definitions to pediatric age groups mostly applied previously validated blood pressure thresholds based on the PELOD-2 development cohort [1,2,3]. Considering the similarities between age-adapted SOFA and PELOD-2, it is thus not surprising that both approaches resulted in comparable performance to predict in-hospital mortality. However, a previous report [2] on the quick PELOD-2 (defined as tachycardia, GCS < 11, and hypotension) suggested a higher performance compared to the age-adapted qSOFA (defined as tachypnea, GCS < 14, and hypotension) in our study [1]. When applying the qPELOD-2 criteria to 2271 children admitted to intensive care with infection, we observed an AUROC of 0.68 (95%-CI 0.63–0.73) for mortality, and of 0.62 (0.60–0.64) for mortality and/or ICU stay of ≥ 3 days, which is similar to our previous analyses on qSOFA indicated a low discriminatory performance of qSOFA in this cohort. These findings demonstrate the importance of independent validation in cohorts outside the population used to derive the score, and question the applicability of qSOFA as a clinically meaningful tool that will improve the care of children with sepsis. Of note, the controversy around best blood pressure thresholds [4] for severity recognition and treatment of pediatric septic shock has left one of the pillars of sepsis management on shaky ground. The need to revise pediatric sepsis definitions [5] may represent an opportunity to combine large cohorts, allowing the reassessment of optimal measures of organ dysfunction in children with sepsis.