Journal Logo


Pediatric Sepsis Biomarker Risk Model With Outcome After PICU Discharge: A Strong Research Tool, but Let Us Not Forget Composite Prognostic Factors!*

Leteurtre, Stéphane MD, PhD; Recher, Morgan MD

Author Information
Pediatric Critical Care Medicine: January 2021 - Volume 22 - Issue 1 - p 125-127
doi: 10.1097/PCC.0000000000002621
  • Free
  • Editor's Choice

Mortality rate of critical illness in children has dropped over the past 20 years, which is a tribute to medical progress! Now, the health target has shifted to a collection of symptoms and problems called postintensive care syndrome (PICS) (1). The framework considering the problems of PICS in pediatrics (PICS-p) involves an evaluation of different domains including development and growth (physical health), school performance (cognitive health), psychologic function (emotional health), family and societal functioning (social health), and trajectory and duration of recovery (2).

In the PICU, sepsis remains a leading cause of morbidity, mortality, and the need for healthcare utilization, worldwide (3). The first study of “Life After Pediatric Sepsis Evaluation” (LAPSE) was carried out in 12 academic PICUs in the United States (n = 389 patients). The study showed that inhospital mortality following community-acquired pediatric septic shock was 9% and that 35% of survivors had significant deterioration from baseline in health-related quality of life (HRQL), which persisted for at least 1 year after hospitalization (4). In addition, factors such as duration of organ dysfunction and need for organ failure rescue (e.g., renal replacement therapy, extracorporeal life support, and cardiopulmonary resuscitation) were associated with death or persistent, serious deterioration of HRQL (PSD-HQRL) at 3 months after PICU admission (5).

Now, in this issue of Pediatric Critical Care Medicine, Wong et al (6) report a secondary analysis of the LAPSE cohort with tests of new biomarkers against three outcomes, long-term mortality, or PSD-HRQL at two times after hospital discharge. The analysis investigated the impact of a “dysregulated host response” as categorized by the “Pediatric Sepsis Biomarker Risk Model” (PERSEVERE) in 173 patients (44% of the original LAPSE cohort). The PERSEVERE biomarkers included C-C chemokine ligand 3 (CCL3), granzyme B (GZMB), heat shock protein 70 kD 1B, interleukin (IL)-8, and matrix metallopeptidase 8 and were measured during the first 24 hours following sepsis shock diagnosis (7). In the present study, the performance of the PERSEVERE biomarker model had limited value in distinguishing between hospital survivors and nonsurvivors (area under the receiver operating characteristic curve [AUROC] 0.73, 95% CI 0.59–0.87). The result may appear to be a little disappointing, but the finding still suggests the importance of biological criteria in pediatric sepsis prognostication. In addition, like any statistical criterion, it is not surprising that this first validation on an external population is weaker than the result obtained in the development or hypothesis-generating population. This corresponds naturally to the necessary adaptation or customization of any prognostic system in a new population with different case-mix profile (8). Of note, here, PERSEVERE-II is an updated version of the biomarker model that incorporates two other predictor variables, admission platelet count, and day 1 acute kidney injury (AKI) state (9); when used to predict severe sepsis-associated AKI on day 3, the AUROC was 0.95 (95% CI, 0.92–0.98).

Wong et al (6) also evaluated PICS-p using mortality or PSD-HQRL at 1 month, and mortality or PSD-HQRL at 3 months in order to develop an updated PERSEVERE model for predicting PSD-HRQL at 3 months in the LAPSE cohort. The new information from this study is that the PERSEVERE biomarkers during emergency life support on the PICU can be used to predict HRQL at 3 months! Indeed, the classification tree had an AUROC of 0.87 (95%CI, 0.80–0.95). As previously mentioned by the authors, one potential weakness of PERSEVERE is that it focused on a relatively short-term outcome, such as mortality (7), but now this deficit has been corrected. However, a limitation of these results, taking into account a new classification tree with the new outcome of PSD-HRQL at 3 months, is the threshold values observed for each of the biomarkers in the PERSEVERE model. For example, even if the initial CCL3 biomarker remains identical in the different classification trees, thresholds for some markers were very different between the PERSEVERE development sample (Supplemental Fig. 1 in [6]) and the PERSEVERE validation sample (Fig. 1 in [6]) (threshold for age from 0.5 yr with GZMB to 11 yr with CCL3, and threshold for IL-8 from 507 to 2,500 pg/mL between the two different PERSEVERE adaptations, respectively [7]). Hence, the clinical understanding of these classification trees remains complex (4).

The generalization of this study may have three main limitations. First, the assays for the PERSEVERE biomarkers may only be available in large university centers (10). However, these biomarkers will gradually become more available as need increases. Second, even though the HRQL criterion is used in large sepsis trials in adults (11) and pediatric populations (12), the equivalence of various domains of quality of life across cultures is debatable and likely very important in pediatric practice (13). Many instruments of quality of life are developed in the English language and need to have cross-cultural adaptation to other languages and cultures (13). Last, as mentioned by Wong et al (6), their study sample remains small for the validation of the PSD-HQRL outcome criterion at 3 months after PICU admission and the results of the PERSEVERE biomarkers must be repeated in more heterogeneous populations.

Overall, some fundamental questions about sepsis, outcome, and prediction tools remain: Why should we trust biomarkers alone compared with a composite clinical tool to assess the prognosis of a child with sepsis, whether in terms of mortality or quality of life? Can a diagnostic or therapeutic decision, or a stratified analysis be based solely on biomarker values? Should we not consider developing a clinical tool taking into account clinical parameters (such as a severity score or a measure of organ dysfunctions) in association with biomarkers (14,15)? Notwithstanding these topics for future research, the performance of informatics, technologic tools, and growing pediatric databases are no longer an obstacle to the construction of complex models linked to outcome such as the assessment of HRQL after PICU in sepsis.


1. Watson RS, Choong K, Colville G, et al. Life after critical illness in children-toward an understanding of pediatric post-intensive care syndrome. J Pediatr. 2018; 198:16–24
2. Manning JC, Pinto NP, Rennick JE, et al. Conceptualizing post intensive care syndrome in children-The PICS-p framework. Pediatr Crit Care Med. 2018; 19:298–300
3. Weiss SL, Peters MJ, Alhazzani W, et al. Surviving sepsis campaign international guidelines for the management of septic shock and sepsis-associated organ dysfunction in children. Pediatr Crit Care Med. 2020; 21:e52–e106
4. Zimmerman JJ, Banks R, Berg RA, et al.Life After Pediatric Sepsis Evaluation (LAPSE) InvestigatorsTrajectory of mortality and health-related quality of life morbidity following community-acquired pediatric septic shock. Crit Care Med. 2020; 48:329–337
5. Zimmerman JJ, Banks R, Berg RA, et al. Critical illness factors associated with long-term mortality and health-related quality of life morbidity following community-acquired pediatric septic shock. Crit Care Med. 2020; 48:319–328
6. Wong HR, Reeder RW, Banks R, et al. Biomarkers for Estimating Risk of Hospital Mortality and Long-Term Quality-of-Life Morbidity After Surviving Pediatric Septic Shock: A Secondary Analysis of the Life After Pediatric Sepsis Evaluation Investigation. Pediatr Crit Care Med. 2021; 22:8–15
7. Wong HR, Salisbury S, Xiao Q, et al. The pediatric sepsis biomarker risk model. Crit Care. 2012; 16:R174
8. Pirracchio R, Ranzani OT. Recalibrating our prediction models in the ICU: Time to move from the abacus to the computer. Intensive Care Med. 2014; 40:438–441
9. Stanski NL, Stenson EK, Cvijanovich NZ, et al. Persevere biomarkers predict severe acute kidney injury and renal recovery in pediatric septic shock. Am J Respir Crit Care Med. 2020; 201:848–855
10. Gien J, Soranno D. Identifying the patient at risk for acute kidney injury: Pediatric sepsis biomarker risk model study. Am J Respir Crit Care Med. 2020; 201:764–766
11. Hammond NE, Finfer SR, Li Q, et al.ADRENAL Trial Investigators and the Australian and New Zealand Intensive Care Society Clinical Trials GroupHealth-related quality of life in survivors of septic shock: 6-month follow-up from the ADRENAL trial. Intensive Care Med. 2020; 46:1696–1706
12. Meert KL, Reeder R, Maddux AB, et al.The Life After Pediatric Sepsis Evaluation (LAPSE) InvestigatorsTrajectories and risk factors for altered physical and psychosocial health-related quality of life after pediatric community-acquired septic shock. Pediatr Crit Care Med. 2020; 21:869–878
13. Tripathy S, Myatra SN. Are the instruments for quality of life assessment comparable between cultures? No. Intensive Care Med. 2020; 46:1746–1748
14. Pollack MM, Holubkov R, Funai T, et al. The pediatric risk of mortality score: update 2015 Pediatr Crit Care Med. 2016; 17:2–9
15. Leclerc F, Duhamel A, Deken V, et al.Groupe Francophone de Réanimation et Urgences Pédiatriques (GFRUP)Can the pediatric logistic organ dysfunction-2 score on day 1 be used in clinical criteria for sepsis in children? Pediatr Crit Care Med. 2017; 18:758–763
Copyright © 2021 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies