This retrospective single-center cohort study of 60 children with sepsis admitted to the PICU from the PED assessed the accuracy of several prognostic scores calculated in the first 24 hours from admission in predicting mortality and morbidity at PICU discharge. A better performance of organ dysfunction scores in predicting death compared to ICCPS-derived criteria (i.e., SIRS criteria and severe sepsis criteria) has been observed. Among the organ dysfunction scores, PELOD-2 presented the best performance measures, resulting significantly more accurate than the ICCPS-derived criteria in discriminating both the primary and some secondary outcomes.
The validity of ICCPS-derived criteria (especially SIRS criteria) to evaluate severity of patients has already been questioned in the last two decades in the adult population, resulting in a paradigmatic change during last Sepsis-3 Consensus Statement [9]. As mentioned before, the consensus resulted in the elimination of the definitions of SIRS and severe sepsis, underlining that sepsis is already differentiated from uncomplicated infection by the presence of a life-threatening organ dysfunction. The operationalization of clinical criteria to identify individuals meeting outcomes consistent with sepsis has conducted to the implementation of the SOFA score. This instrument, validated using big but limited-to-adult datasets [10], is now considered the gold standard to prognosticate mortality in adult patients with suspected infection.
Recently, several authors compared the performance of pediatric adapted organ-dysfunction scores (i.e., pSOFA, PELOD-2) as outcome predictors in large cohort of critically ill children, resulting in excellent performance of organ dysfunction score in predicting mortality and other relevant outcomes [12; 16]. Our study confirms these recent new evidences. In our study, we did not find any clear statistical superiority of an organ-dysfunction score compared to the others, probably secondary to the small dimension of the sample size. However, as already stated, PELOD-2 demonstrated the highest performance at discriminating mortality, supporting findings of other studies [16], and suggesting its promising use to standardize definitions and diagnostic criteria for pediatric sepsis.
A limitation of previous studies is that they did not evaluate the prognostic accuracy of organ-dysfunction scores in any functional outcome of the patients. These outcomes are becoming the reference points in the short and long-term evaluation of patients after PICU admission, especially considering recent improvements on sepsis mortality rate over the last two decades [2]. The POPC [21] is a qualitative tool validated for assessing functional morbidity in large cohorts of critically-ill children [26; 27]. In multiple retrospective studies by Typpo and colleagues [28; 29], the presence of day-1 multiple organ dysfunction (at the time quantified trough ICCPS criteria) was significantly associated with death or change in POPC score greater than 3 points compared to baseline. In our cohort, about half of the patients had at least mild disability at PICU admission, which is in agreement with data reported by other authors on previous large cohort studies [29; 30]. At PICU discharge, 12% of the patients presented a new disability, while 18.3% of the patients had a “poor outcome”, i.e. a composite outcome of mortality and new disability [20]. PELOD-2 resulted as the organ-dysfunction score showing the best performance in predicting a "poor outcome”, although being significantly higher only compared to severe sepsis score. These preliminary results need to be replicated in larger populations.
Interestingly, almost 97% of our patients resulted classified as “severe sepsis” by ICCPS criteria. This data suggests a high severity of the patients in our cohort, likely reflecting a high institutional threshold for admitting patients in the PICU from the PED. This is consistent with a median PELOD-2 score of 7 (IQR 6.0–11.0) which is higher compared to ones reported in other previous cohorts [16]. We consider that a limitation, as probably a percentage of septic patients with more favorable outcomes were managed outside the PICU, limiting the variability and the dimension of the sample included in our study.
Overall, these findings support a trend of recent evidence in the pediatric critical care community that suggests a central role for organ-dysfunction scores to help standardize prognostication in pediatric sepsis. In particular, this study is among the first ones that showed a better prognostic accuracy of organ-dysfunction scores in predicting mortality and with promising results also on predicting the morbidity (e.g., new disability) of patients with sepsis.
The present study presents several limitations. First, the results were generated using retrospective data from a single center. However, most of the outcome data were retrieved from a prospective compiled registry, limiting the number of missing data in the sample. Second, we did not consider a comprehensive cohort of patients with infection but only patients with a confirmed diagnosis of sepsis, according to ICCPS criteria, limiting our chance to make assumption on the performance of ICCPS criteria from a diagnostic standpoint. Third, only patients admitted from the PED have been assessed and only the first 24 hours of hospitalization of the patient (comprehensive of the time in the PED) have been considered for the analysis. This limited the study sample to subjects with community-acquired infections and our prognostic considerations to this particular timespan of patients’ care. However, this aspect could be considered also a point of strength, as the aim was to evaluate the prognostic accuracy of multiple tools in patients with sepsis within the first hours of hospital admission. Fourth, the relatively small sample size might have hindered the analysis, in particular resulting in a less precise estimation of the accuracy of the scores. These limitations reduce the generalization of the present findings highlighting the need for future prospective, multicenter, larger studies to draw firmer conclusions.