Pediatric Sepsis Diagnosis, Management, and Sub-phenotypes

Abstract

Sepsis and septic shock are major causes of morbidity, mortality, and healthcare costs for children worldwide, including >3 million deaths annually and, among survivors, risk for new or worsening functional impairments, including reduced quality of life, new respiratory, nutritional, or technological assistance, and recurrent severe infections. Advances in the understanding of sepsis pathophysiology highlight a need to update the definition and diagnostic criteria for pediatric sepsis and septic shock, while new data support an increasing role for automated screening algorithms and biomarker combinations to assist earlier recognition. Once sepsis or septic shock is suspected, attention to prompt initiation of broad-spectrum empiric antimicrobial therapy, fluid resuscitation, and vasoactive medications remain key components to initial management with several new and ongoing studies offering new insights into how to optimize this approach. Ultimately, a key goal is for screening to encompass as many children at risk for sepsis and trigger early treatment without increasing unnecessary broad-spectrum antibiotics and preventable hospitalizations. While the role for adjunctive treatment with corticosteroids and other metabolic therapies remains incompletely defined, ongoing studies will soon offer updated guidance for optimal use. Finally, we are increasingly moving toward an era in which precision therapeutics will bring novel strategies to improve outcomes, especially for the subset of children with sepsis-induced multiple organ dysfunction syndrome and sepsis sub-phenotypes for whom antibiotics, fluid, vasoactive medications, and supportive care remain insufficient.

Article Summary

This article summarizes recent evidence-based practices for diagnosis and initial management of children with sepsis, including defined sepsis sub-phenotypes linked to precision therapeutics.

Introduction

Sepsis and septic shock are major causes of morbidity, mortality, and healthcare costs for children worldwide.^{1, 2} In 2017, an estimated 20 million cases of sepsis occurred in children <5 years and 5 million cases in those 5 to <19 years, resulting in 3.5 million deaths.³ Children account for half of all global sepsis cases and one-third of deaths with a disproportionate burden in Africa, South America, and the Indian subcontinent.^{3, 4} Case-fatality rates vary from 3–5% for sepsis in previously healthy children to as high as 25–50% for sepsis in children with comorbid medical conditions, hospital-acquired infections, limited access to healthcare, and malnutrition.^{2, 5} Almost all pediatric sepsis-related deaths involve multiple organ dysfunction syndrome (MODS), defined as dysfunction in ≥2 organ systems.⁶ Early deaths (within 72 hours) are most often due to acute MODS with shock/cardiac arrest, while later deaths are driven by persistent MODS and withdrawal of life-sustaining technological support.^6–8 Among recent studies of children who survive treatment for sepsis requiring intensive care, up to one-third have sustained new or worsening functional impairments, including reduced quality of life, new respiratory, nutritional, or technological assistance, and recurrent severe infections.^9–14

The objective of this report is to summarize recent evidence-based practices for the diagnosis and initial management of children with sepsis, including the increasing emphasis on sepsis sub-phenotypes linked to precision therapeutics (Table 1). Application of optimal general and targeted therapies can reduce sepsis-associated morbidity and mortality in children overall, while identification of sub-phenotypes with measurable pathophysiologic abnormalities will increasingly facilitate targeted adjunctive therapies for children in whom antibiotics, fluid, vasoactive medications, and supportive care remain insufficient.

Table 1:

Summary of State-of-the-Art Findings for the Diagnosis and Management of Pediatric Sepsis and Septic Shock

Focus Area	State-of-the-Art Key Points	Knowledge Gaps
Diagnosis of Sepsis
Defining sepsis	Sepsis is better conceived of as life-threatening organ dysfunction caused by a dysregulated host response to infection rather than SIRS Sepsis pathophysiology involves both pro- and anti-inflammatory mediators superimposed on genomic, proteomic, and metabolomic reprogramming of the immuno-inflammatory response	The optimal criteria to define and operationalize organ dysfunction and the dysregulated host response in children with severe infections is an area of active investigation
Diagnosing sepsis	Clinicians should use abnormal vital signs, laboratory evidence of inflammation, clinical signs of infection, and skin perfusion to recognize sepsis and septic shock in children even if these criteria have not yet met published thresholds to define sepsis Biomarkers, such as lactate, PCT, and CRP, can be used to assist diagnosis and management, but are not “gold-standard” tests	Combinations of biomarkers representing various pathophysiologic mechanisms, including the immunoinflammatory response, endotheliopathy, microvascular/glycocalyx degradation, and intestinal permeability, may offer improved accuracy for early diagnosis, risk stratification, and monitoring response to therapy
Screening for sepsis	Systematic screening is associated with improved outcomes and, therefore, should be implemented to enhance timely recognition of septic shock and other sepsis-associated organ dysfunction in children who preset as acutely unwell	Current screening algorithms that relay on vital signs, nursing assessments, and laboratory values have low PPV AI algorithms that incorporate high-fidelity physiologic data may shorten time to sepsis recognition and improve accuracy, but impact on workflow and outcomes is unknown
Initial Management of Pediatric Sepsis
Empiric broad-spectrum antibiotics	For children with septic shock, antimicrobial therapy should be administered as soon as possible, ideally within 1 hour of recognition For children with suspected sepsis without shock, an expedited diagnostic examination to confirm sepsis should be completed, with antibiotics for those most likely to benefit given as soon as possible but not later than 3 hours A bundled approach in which early antibiotics are administered along with other therapies provides maximal benefit	Because independent benefit of antibiotics administered within 1 hour remains unclear, the utility to use a dichotomous 1 hour cut-point to classify high- versus low-quality care is not konwn
Fluid resuscitation	Fluid bolus therapy can increase cardiac output and tissue perfusion in patients with hypovolemia, but the benefit is often transient Fluid bolus therapy in 10–20 mL/kg aliquots administered over 5–20 minutes is recommended for children with septic shock treated in healthcare systems with availability of intensive care (either locally or via transport) over the first hour of resuscitation and titrated to clinical markers of cardiac output Fluid bolus therapy is not recommended for children with sepsis or septic shock treated in healthcare systems without access to mechanical ventilation, advanced hemodynamic monitoring, or sustained vasoactive and/or mechanical circulatory unless the child exhibits hypotension Crystalloids, possibly balanced/buffered solutions, are preferred as first-line option for fluid bolus therapy Dynamic tests of fluid responsiveness and measures of cardiac output should guide fluid bolus therapy and help reduce harm from fluid overload	The optimal volume and timing of fluid bolus therapy is not known but likely varies across individual patients The comparative effectiveness of 0.9% saline versus balanced/buffered fluids is unknown and an active area of study The utility of early administration of vasoactive infusions to limit or replace fluid bolus therapy is unknown and an active area of study
Vasoactive medications	Epinephrine or norepinephrine are preferred over dopamine to reverse shock and prevent mortality Epinephrine is preferred for patients with moderate-severe myocardial dysfunction Norepinephrine is preferred for patients with vasoplegia	The comparative effectiveness of epinephrine versus norepinephrine as first-line vasoactive support is unknown The utility of adjunctive vasoactive medications, (milrinone, vasopressin, angiotensin II), is unknown
Corticosteroids and metabolic resuscitation	Hydrocortisone is not recommended for children with sepsis or septic shock who are hemodynamically stable following fluid resuscitation and low-dose vasoactive support There are insufficient data to support routine adjunctive therapy with HAT (hydrocortisone, ascorbic acid, and thiamine)	The utility of hydrocortisone (with or without fludrocortisone) to improve outcomes for children who require moderate- to high-dose vasoactive support is unknown and an area of active study The utility of metabolic therapies to restore tissue oxygen utilization and cellular/metabolic homeostasis is unknown
Sepsis Sub-phenotypes and Precision Therapies
	Multiple sepsis sub-phenotypes have been linked to biologic mechanisms for which targeted therapies are available	The optimal method to detect sepsis sub-phenotypes in clinical practice and the utility of targeted therapies that rely on prognostic and predictive enrichment are unknown and an area of active study

Diagnosis of Pediatric Sepsis

Defining Sepsis

Through the early 2000’s, the prevailing characterization of sepsis was that of an infection-induced systemic inflammatory response syndrome (SIRS) that led to shock and organ dysfunction often remote from the site of infection. More recently, understanding of sepsis pathophysiology has advanced beyond that of a primary hyper-inflammatory state (or “cytokine storm”) to a condition in which both pro- and anti-inflammatory mediators are superimposed on genomic, proteomic, and metabolomic reprogramming of the entire immuno-inflammatory response.^{15, 16} As a result, in 2016, Sepsis-3 updated the definition of sepsis for adult patients as “life-threatening organ dysfunction caused by a dysregulated host response to infection” and septic shock as the most severe state in which “profound circulatory, cellular, and metabolic abnormalities are associated with a greater risk of mortality than with sepsis alone”.¹⁷

For children, the most recent and commonly cited definition of sepsis remains the 2005 International Pediatric Sepsis Consensus Conference (IPSCC), which relies on the presence of SIRS.^{4, 18} However, there is increasing recognition that sepsis in children, as in adults, is better conceptualized as a dysregulated host response to infection that leads to life-threatening organ dysfunction.^{19, 20} In a study of 1,092 children treated for sepsis in a pediatric intensive care unit in China, the presence of organ dysfunction (area under the receiver operator curve [AUROC] 0.74, 95% CI 0.71–0.77) was more predictive of sepsis-related death than SIRS (AUROC 0.56, 95% CI 0.51–0.60).²¹ Similarly, among 2,594 patients admitted with infection to PICUs in Australia and New Zealand, SIRS had lower predictive validity for mortality than organ dysfunction scores.²²

Currently, there are several criteria to define organ dysfunction in pediatric sepsis (Table 2).^23–29 Most recently, the Pediatric Organ Dysfunction Information Update Mandate (PODIUM) developed an updated set of dichotomous diagnostic criteria for dysfunction in nine organ systems using literature review and expert consensus.²⁴ Among 22,427 undifferentiated PICU patients, PODIUM was able to discriminate hospital mortality as well or better than IPSCC criteria, though validation in pediatric sepsis is not yet available.²⁸ Efforts to identify the most useful organ dysfunction criteria are ongoing through an international taskforce working to update definitions for pediatric sepsis and septic shock.³⁰

Table 2:

Criteria for Organ Dysfunction in Pediatric Sepsis

Organ System	Proulx et al, 1996	IPSCC, 2005	pSOFA, 2017
Cardiovascular	Systolic BP <40 mmHg for age <12 months or <50 mm Hg for age ≥12 months OR Heart rate <50 or >220 beats/min for age <12 months or <40 or >200 beats/min for age ≥12 months OR Cardiac arrest OR Serum pH <7.2 with normal PaC02 OR Need for vasoactive drug to maintain BP and/or cardiac output (dopamine ≤5 pg/kg/min excluded)	Decrease in BP<5th percentile for age or systolic BP <2 SD below normal for age ORNeed for vasoactive drug to maintain BP in normal range ORTwo of the following: • Unexplained metabolic acidosis: base deficit >5.0 mEq/L • Increased arterial lactate >2 times upper limit of normal • Oliguria: urine output <0.5 mL/kg/hr • Prolonged capillary refill: >5 secs • Core to peripheral temperature gap >3°C	0: Normal MAP for age 1: Low MAP for age 2: Dopamine/dobutamine ≤5 mcg/kg/min 3: Dopamine/dobutamine >5 mcg/kg/min or epinephrine/ norepinephrine ≤0.1 mcg/kg/min 4: Dopamine/dobutamine >15 mcg/kg/min or epinephrine/ norepinephrine >0.1 mcg/kg/min
Respiratory	RR >90 breaths/min for age <12 months or >70 breaths/min for age ≥12 months OR PaC02 >65 mm Hg (>8.7 kPa) OR Pa02 <40 mm Hg (<5.3 kPa) in absence of cyanotic congenital heart disease OR PaO2/FIO2 <200 in the absence of cyanotic congenital heart disease OR Invasive mechanical ventilation (for >24 h in a postoperative patient)	PaO2/FIO2, 300 in absence of cyanotic heart disease or preexisting lung disease OR PaCO2 >65 torr or 20 mm Hg over baseline PaCO2 OR Proven need for >50% FIO2 to maintain saturation ≥92% OR Need for non-elective invasive or noninvasive mechanical ventilation	0: PaO2:FiO2 ≥400 or SpO2:FiO2 ≥292 1: PaO2:FiO2 300–399 or SpO2:FiO2 265–291 2: PaO2:FiO2 200–299 or SpO2:FiO2 221–264 3: PaO2:FiO2 100–199 or SpO2:FiO2 148–220 with respiratory support 4: PaO2:FiO2 <100 or SpO2:FiO2 <148 with respiratory support
Renal	Blood urea nitrogen >100 mg/dL (>36 mmol/L) OR Serum creatinine ≥2 mg/dL (≥177 μmol/L in the absence of pre-existing renal disease OR Dialysis	Serum creatinine ≥2 times upper limit of normal for age or 2-fold increase in baseline creatinine	0: normal serum creatinine for age 1: minimally elevated serum creatinine for age 2: mildly elevated serum creatine for age 3: moderately elevated serum creatinine forage 4: severely elevated serum creatinine for age
Hepatic	Total bilirubin level >3 mg/dL excluding icterus due to breast feeding	Total bilirubin ≥4 mg/dL (not applicable for newborn) OR ALT 2 times upper limit of normal for age	0: Bilirubin <1.2 mg/dL 1: Bilirubin 1.2–1.9 mg/dL 2: Bilirubin 2.0–5.9 mg/dL 3: Bilirubin 6.0–11.9 mg/dL 4: Bilirubin >12 mg/dL
Hematologic	Hemoglobin < 5 g/dL (<50 g/L) OR WBC count < 3,000 cells/mm3 OR Platelet <20,000 cells/mm3 OR D-dimer >0.5 μg/mL with prothrombin time >20 sec or partial thromboplastin time >60 sec	Platelet count <80,000/mm3 or a decline of 50% in platelet count from highest value recorded over the past 3 days OR International normalized ratio >2	0: Platelet ≥150 ×103/µL 1: Platelet 100–149 ×103/µL 2: Platelet 50–99 ×103/µL 3: Platelet 20–49 ×103/µL dL 4: Platelet <20 ×103/µL
Neurologic	GCS <5 OR Fixed dilated pupils	GCS ≤11 OR Acute change in mental status with a decrease in GCS ≥3 points from abnormal baseline	0: GCS 15 1: GCS 13–14 2: GCS 10–12 3: GCS 6–9 4: GCS <6
Gastrointestinal	Gastroduodenal bleeding causing one of the following: • Decrease in hemoglobin of ≥2 g/dL (≥20 g/L) in 24 h • Blood transfusion • Hypotension with BP <3rd percentile for age • Gastric or duodenal surgery • Death	Not included

MAP, mean arterial pressure; BP, blood pressure; SD, standard deviation; RR, respiratory rate; PaCO2, partial pressure of carbon dioxide in arterial blood; PaO2, partial pressure of oxygen in arterial blood; FIO2, fraction of inspired air that is oxygen; AST, aspartate aminotransferase; LDH, lactate dehydrogenase; GCS, Glasgow Coma Score

Diagnosing Sepsis

Importantly, the criteria used at the bedside to diagnose and treat suspected sepsis do not need to perfectly align with operational cut-points that define sepsis. Clinicians can—and should—consider criteria to recognize and diagnose children with suspected sepsis or septic shock that are timely and support sepsis over alternative diagnoses. An international survey of 2,835 clinicians from pediatric intensive care, emergency medicine, and infectious diseases reported that abnormal vital signs, laboratory evidence of inflammation, clinical signs of infection, and skin perfusion were more useful to recognize sepsis and septic shock in children in clinical practice than strict adherence to 2005 IPSCC or Sepsis-3 definitions.²⁰ This approach acknowledges that sepsis lays somewhere in the evolution from uncomplicated infection to death without a clear delineation. Therefore, attention to early and subtle alterations in temperature (high or low), tachycardia, tachypnea, altered mental status, delayed or flash capillary refill, cool distal extremities, diminished or bounding peripheral pulses, decreased urine output, low blood pressure, and narrow or wide pulse pressure can indicate that a child should be treated for sepsis/septic shock even before crossing an arbitrary threshold.

There has been longstanding interest to identify biomarkers to assist in the diagnosis of sepsis, with blood lactate, C-reactive protein (CRP), and procalcitonin (PCT) the most commonly used.³¹ Blood lactate levels rise when oxygen delivery to tissues is insufficient to support metabolic demand, resulting in anaerobic conversion of pyruvate to lactate rather than driving adenosine triphosphate (ATP) production through mitochondrial aerobic respiration.³² With reversal of shock to restore oxygen and substrate delivery to tissues, blood lactate levels typically normalize. However, there are many other causes of hyperlactatemia (including artifactual increases from difficult phlebotomy), and septic shock can occur prior to or even without a rise in blood lactate. As a result, blood lactate is less useful to stratify acutely unwell children for risk for sepsis, but should be trended over time as one method to monitor response to therapy after sepsis is suspected.^{33, 34}

CRP is a hepatic-derived acute phase reactant that serves as a sensitive, albeit non-specific, marker of inflammation, and PCT is a pro-hormone of calcitonin ubiquitously secreted in response to bacterial-mediated cytokines. PCT offers a more specific, and generally earlier, biomarker for invasive bacterial infections than CRP.³⁵ Both PCT and CRP are useful for differentiating groups of patients with sepsis from uncomplicated infection and have high negative predictive values to exclude severe illness from sepsis, but their clinical utility within individuals is tempered by frequent false-positive results and the potential for erroneous reassurance if measured very early in the course of an evolving illness.³⁶

Several other biomarkers have demonstrated some utility to help diagnosis and/or risk-stratify patients with sepsis, including presepsin, sTREM, seruma amyloid A, sCD163, pentraxin-3, IL-7, and IL-8, microRNAs, among others.^{31, 37} It is unlikely, however, that any one biomarker could represent the totality of the complex biology of sepsis. Instead, measuring several biomarkers concurrently has shown promise to overcome the limitations of any single biomarker, as has combining differential expression of multiple genes.³⁸ Increasing data also suggests that select biomarkers can offer insight into different aspects of sepsis pathophysiology beyond the immunoinflammatory response. For example, biomarkers of endothelial activation (e.g., angiopoietin-1/2, soluble fms-like tyrosine kinase-1, thrombomodulin, vascular endothelial growth factor), microvascular dysfunction and glycocalyx degradation (e.g., perfused boundary region on sublingual video microscopy, syndecan-1), and intestinal permeability (e.g., claudin-3, intestinal fatty acid binding protein, citrulline) are active areas of research to enhance recognition of high-risk groups.^39–42 Less invasive volatile organic compounds within exhaled breath have also demonstrated early promise to enhance detection of infections associated with sepsis.⁴³ Additional evidence of impact on patient outcomes along with availability of testing with rapid reporting of results is necessary before these can be recommended for widespread utilization, but it is very likely that such tools will become increasingly available for clinical implementation in the future.

Screening for Sepsis

Systematic screening is recommended to assist with timely recognition of septic shock and other sepsis-associated organ dysfunction in children who preset as acutely unwell.^{33, 34} A systematic process to trigger deliberate consideration that an unwell children may be suffering from sepsis reduces time to treatment, shortens length of stay, and decreases missed sepsis diagnoses.^44–46 Electronic algorithms that continuously measure perturbations in vital signs, nursing assessments, and laboratory values are generally more sensitive and specific than appraisals at pre-determined time intervals, but the reported range of positive predictive value (PPV) remains only between 4% and 49%.⁴⁷ Screening algorithms that incorporate measures of illness severity and account for comorbid conditions are helpful to improve accuracy but continue to rely on clinician-directed inputs into the electronic health record. For example, the commonly used Pediatric Septic Shock Collaborative trigger tool that combines vital signs, physical exam findings, and comorbid conditions has demonstrated 99% sensitivity and 20% PPV,⁴⁷ while an alternative tool based on SIRS and IPSCC organ dysfunction criteria without consideration of comorbid conditions achieved only 72% sensitivity and 8.1% PPV.⁴⁵ To overcome the limitations of low PPV and reduce alert fatigue, a commonly used approach has been a multi-stage identification method whereby an initial automated alert triggers a rapid nursing assessment. If the nursing assessment corroborates possible sepsis, a huddle is then called to decide whether to proceed treatment for sepsis.⁴⁴

Artificial intelligence (AI) algorithms that incorporate high-fidelity physiologic data directly from bedside monitors have the potential to identify shock physiology hours before overt changes in vital signs and improve overall accuracy.⁴⁸ For example, in a study of 493 children admitted to one PICU, machine learning algorithms that used standard deviation of heart rate, blood pressure, and symbolic transitions probabilities of those variables identified sepsis up to eight hours earlier than a screening tool that relied on traditional SIRS and organ dysfunction criteria.⁴⁹ However, AI-based algorithms still need to better incorporate physiologic variation from comorbid illnesses to truly improve upon EHR-based tools, and the impact of such real-time, AI-driven detection tools on provider workflow and patient outcomes remains to be tested.

Initial Management of Pediatric Sepsis

The Surviving Sepsis Campaign recommends six key initial management steps for children with suspected sepsis or septic shock: 1) obtain intravenous/intraosseous access, 2) collect a blood culture (and other diagnostic tests for most likely sites of infection), 3) start empiric broad-spectrum antibiotics, 4) measure blood lactate, 5) administer fluid bolus(es) if shock is present, and 6) start vasoactive agents if shock persists.^{33, 34} These interventions should be completed as quickly as possible—ideally within one hour of initial recognition for children with septic shock and within three hours of initial suspicion of sepsis in the absence of shock. The Improving Pediatric Sepsis Outcomes collaborative found that adherence to a bundle that included use of a sepsis recognition method, first fluid bolus within 60 minutes, and antibiotic administration within 180 minutes was associated with reduced mortality.⁵⁰ Treatment of hypoxemia with supplemental oxygen (goal SpO₂ 92–98%) and reversal of hypoglycemia and ionized hypocalcemia are also important. Patients should be continuously reassessed for signs of ongoing shock, evolving fluid status (including pulmonary rales, hepatomegaly, peripheral edema, and echocardiographic findings that may suggest fluid overload), and response to each intervention while arranging for transfer to an appropriate intensive/critical care unit.

Empiric Broad-Spectrum Antibiotics

Bacteria are the primary etiology of sepsis in children in up to 78% of cases.^{8, 51} In addition, severe viral illnesses, including influenza and SARS-CoV-2, may be accompanied by concurrent bacterial infections.^{52, 53} These data offer strong biologic rationale for rapid administration of empiric antibiotic therapy targeting all likely pathogens. The initial choice of antimicrobials should be guided by the patient’s history and symptoms, along with local microbial epidemiology. In most cases, a single broad-spectrum antibiotic is sufficient, such as a third-generation cephalosporin (e.g., ceftriaxone) in previously healthy children with community-acquired sepsis. For children with immunocompromising conditions or in hospital-acquired sepsis, an antibiotic with extended anti-pseudomonal coverage should be used.^{33, 34} Additional antimicrobials should be included to extend the range of coverage for patients at risk of drug-resistant pathogens (e.g., methicillin-resistant staphylococcus aureus) or to limit toxin production with toxic shock syndrome and or necrotizing fasciitis. Antiviral drugs should also be included in select cases, such as neuraminidase inhibitors for influenza or acyclovir for herpes simplex virus (especially in neonates).

For children with septic shock, antimicrobial therapy should be administered as soon as possible, ideally within one hour of recognition.^{33, 34} Several studies have demonstrated improved outcomes for children with sepsis/septic shock when rapid antibiotics (often within one hour of recognition) are included as part of a bundle of care, but the independent impact of antibiotic timing is less clear. One observational study of 130 children from a single PICU, most of whom presented with shock, reported a progressive rise in mortality with each hour delay in antibiotic administration from sepsis recognition, but this association only became significant after a delay of >3 hours (OR 3.92, 95% CI 0.1.27–12.06).⁵⁴ Nonetheless, based on data pointing towards a likely benefit of early antibiotic therapy for septic shock, current standard of care should include empiric antimicrobial therapy as soon as possible, with 1-hour being a reasonable goal. However, to avoid hazards of unnecessary antibiotic exposure and because mortality risk does not appear to increase significantly within the first three hours, it is reasonable to first conduct an expedited diagnostic examination to confirm sepsis and need for antibiotics for those children who do not present with shock. Once sepsis-induced organ dysfunction is confirmed, rapid microbiological testing supports a bacterial infection, or the patient’s condition evolves to include shock, antibiotics should be promptly administered.

Fluid Resuscitation

Fluid bolus therapy seeks to restore intravascular volume hypovolemia in septic shock resulting from reduced intake, post-capillary venule leak and low plasma oncotic pressure (“third-spacing”), and increased fluid losses (such as from fever or diarrheal illness). Through the Frank-Starling principle, a fluid bolus can increase venous return, augment ventricular filling, and induce stretch of cardiac myocytes that often translates to increased contractility, higher stroke volume, and improved microcirculatory blood flow and tissue perfusion. In a fluid-responsive patient without severe sepsis-induced myocardial dysfunction, a fluid bolus of 20 mL/kg will increase stroke volume by approximately 10–20%.⁵⁵ Unfortunately, this hemodynamic response is often transient. For example, in a recent study of 41 children presenting with septic shock to an emergency department in Australia, the median improvement in cardiac index immediately after a fluid bolus was 18%, but only four children sustained this benefit for at least 60 minutes.⁵⁵ Such transient hemodynamic response often leads to repeated administration of fluid boluses. While this practice may be necessary to reverse the hypovolemic component of septic shock, it also risks exacerbating tissue ischemia though progressive interstitial edema and fluid overload.⁵⁶

Fluid bolus therapy in 10–20 mL/kg aliquots is recommended for children with septic shock treated in healthcare systems with availability of intensive care (either locally or via transport) over the first hour of resuscitation and titrated to clinical markers of cardiac output.^{33, 34} Fluid boluses should be administered over 5–20 minutes, as faster administration times have been associated with adverse respiratory sequelae and slower administration less often improves stroke volume.^{57, 58} Clinical markers of cardiac output may include heart rate, blood pressure, capillary refill time, level of consciousness, extremity temperature, urine output, blood lactate, and central venous oxygen saturation. Because overzealous fluid administration is associated with harm, fluid bolus therapy should be carefully titrated and discontinued if signs of fluid overload develop or no further hemodynamic improvement is observed.

In contrast, for healthcare systems in geographic regions where intensive care is not accessible either locally or via transport, fluid bolus therapy should be avoided unless the child exhibits hypotension.^{33, 34} A lack of accessible intensive care means the inability to deliver interventions such as mechanical ventilation, advanced hemodynamic monitoring, or sustained vasoactive and/or mechanical circulatory support regardless of whether these interventions occur within a formal intensive/critical care unit. These diverging recommendations reflect disparate evidence from studies across different geographic regions. Numerous observational studies, mostly from North America and Europe, report improved outcomes for children who received ≥40 mL/kg in the first hour of resuscitation.^{59, 60} Conversely, the Fluid Expansion as Supportive Therapy (FEAST) study of 3,141 children with severe febrile illness treated in low-resource settings in Africa demonstrated lower mortality (RR 0.72, 95% CI 0.57–0.90) after 48 hours in the group randomized to conservative (i.e., no bolus fluid, maintenance fluid only) rather than liberal fluid therapy (i.e., 20 mL/kg fluid bolus followed by maintenance fluid).⁶¹

It is not known if early administration of vasoactive medications in place of ongoing fluid bolus therapy is beneficial. In a recent open-labeled trial of 63 pediatric patients treated for septic shock in Paraguay, mortality, need for mechanical ventilation, and persistent shock at one hour were decreased with initiation of epinephrine after 40 ml/kg fluid bolus rather than additional fluid boluses.⁶² However, illness severity was higher in the group treated with more liberal fluid boluses. Moreover, two recent randomized clinical trials in adult septic shock failed to demonstrate that intravenous fluid restriction improved outcomes compared to standard intravenous fluid therapy.^{63, 64} The Squeeze study (www.clinicaltrials.gov/NCT03080038) is testing whether septic shock reversal is faster with randomization to an early goal-directed fluid-sparing resuscitation strategy versus usual care in children.

To balance the risks and benefits of fluid bolus therapy, various methods have tried to predict fluid responsiveness, i.e., which patients will increase cardiac output in response to fluid bolus therapy. In general, dynamic measures, such as passive leg raise or gentle hepatic compression, more accurately predict fluid responsiveness than static measures, such as heart rate or central venous pressure.⁶⁵ There is also increasing interest in cardiac ultrasound to predict fluid responsiveness by measuring respiratory variation in inferior vena cava diameter, aortic blood flow peak velocity, and aortic velocity-time integral. However, the accuracy and reproducibility of these measurements to guide fluid therapy is an ongoing challenge, especially in spontaneously breathing children without invasive mechanical ventiation.^{65, 66}

Crystalloid solutions are generally preferred for the initial provision of fluid bolus therapy because crystalloids are inexpensive, readily available, highly compatible with other medications, and there is not clear evidence that outcomes in undifferentiated shock are improved with early use of colloid solutions, such as albumin.^{33, 34} Historically, 0.9% saline has been the most common crystalloid used for fluid resuscitation, largely due to its isotonicity with human plasma. However, increasing evidence supports that the supra-physiologic chloride concentration in 0.9% saline can induce metabolic acidosis, decrease renal blood flow, promote inflammation, and increase risk for endothelial/glycocalyx dysfunction.⁶⁷ In contrast, balanced/buffered fluids, such as Ringer’s lactate, Hartmann’s solution, and PlasmaLyte, have been associated with lower rates of hyperchloremia, acute kidney injury, need for renal replacement therapies, and death in several pediatric studies.^{68, 69} A recent meta-analysis that included four large interventional trials published in the last five years also reported a small, but not statistically significant, reduction in mortality in critically ill adults (risk ratio 0.93, 95% CI 0.76–1.15).⁷⁰ The Pragmatic Pediatric Trial of Balanced versus Normal Saline Fluid in Sepsis (PRoMPT BOLUS) study is currently enrolling 8,800 children with suspected septic shock to test the comparative effectiveness of initial resuscitation with balanced/buffered fluids compared to 0.9% saline (www.clinicaltrials.gov NCT04102371).⁷¹

Vasoactive Medications

Vasoactive medications are useful to treat myocardial dysfunction and manipulate systemic vascular resistance for children with fluid-refractory septic shock. Epinephrine and norepinephrine, rather than dopamine, are preferred as the initial catecholaminergic agents for children who continue to have abnormal perfusion after 40–60 mL/kg of fluid resuscitation, or sooner if myocardial dysfunction or fluid overload is evident.^{33, 34} Two recent randomized clinical trials, with a combined total of 160 children with septic shock, demonstrated faster resolution of shock and improved survival with epinephrine compared to dopamine.^{72, 73} However, no studies have directly compared epinephrine to norepinephrine in children with septic shock. Both epinephrine and norepinephrine target myocardial β1 adrenergic receptors that increase chronotropy and inotropy and vascular α1 adrenergic receptors that increase systemic vascular resistance. Epinephrine at low doses (usually <0.1 µg/kg/min) offers higher affinity for vascular β2 receptors that decrease systemic vascular resistance and promote vasodilation, while norepinephrine only has vasoconstrictor effects. Consequently, epinephrine is preferred to treat children with sepsis-induced myocardial dysfunction and low cardiac output, which is the more common clinical presentation of septic shock in infants and young school-aged children with community-acquired sepsis.^{33, 34} Norepinephrine is preferred to increase systemic vascular resistance for those with preserved or only mild myocardial dysfunction who present with vasoplegia, which is more typical of adolescent and hospital-acquired sepsis.⁷⁴

Other vasoactive medications have been insufficiently tested in pediatric septic shock to be recommended for initial therapy, but dobutamine, milrinone, and vasopressin are commonly used as secondary or adjunctive agents along with advanced hemodynamic monitoring. Recent interest has also focused on angiotensin II, a naturally occurring hormone with marked vasoconstrictor effects triggered through activation of the renin-angiotensin system. Angiotensin II was recently approved for refractory adult shock after a study of 344 adults with non-specific vasodilatory shock found improved likelihood of increased mean arterial blood pressure with angiotensin II added to norepinephrine compared to the addition of placebo.⁷⁵ However, the reported use of angiotensin II for pediatric shock has been thus far limited to small reports.⁷⁶

Corticosteroids and Metabolic Resuscitation

The optimal use of corticosteroids—including timing, type, and dose—for septic shock continues to be incompletely defined. Observational studies in pediatric sepsis have reported an association of corticosteroid use with worse outcomes, but are limited by preferential use of corticosteroids in patients with greater illness severity.⁵¹ At least one pediatric trial suggested that adjunctive corticosteroids hasten resolution of shock.⁷⁷ Of the four high-quality contemporary clinical trials in adults, two reported a mortality reduction with low-dose hydrocortisone (both of which included fludrocortisone) and two did not.^78–80 A recent meta-analysis of 42 clinical trials including 9,969 adults and 225 children with sepsis found that corticosteroid use possibly results in a small reduction in mortality.⁸¹ Consequently, current guidelines recommend against routine use of intravenous hydrocortisone in children if hemodynamic stability is restored with fluid and low-dose vasoactive medications but acknowledges equipoise for children with fluid-refractory, catecholamine-resistant shock.^{33, 34} The Stress Hydrocortisone in Pediatric Septic Shock (SHIPSS) study is actively testing whether adjunctive hydrocortisone for children with fluid and vasoactive-dependent septic shock improves mortality and health-related quality of life (www.clincialtrials.gov/NCT03401398). However, there are no data on the addition of fludrocortisone to enhance the mineralocorticoid effect of corticosteroid administration in pediatric septic shock.

There is also increasing interest in the role of adjunctive metabolic resuscitation strategies to restore tissue oxygen utilization and cellular/metabolic homeostasis that could hasten recovery of immune and organ system function during sepsis. Most recently, interest in the combination of hydrocortisone, high-dose ascorbic acid (vitamin C), and thiamine (vitamin B1)— “HAT” therapy—was ignited by a report of a reduction in mortality among adults with septic shock treated with HAT compared to historical controls from a single center.⁸² A propensity-matched analysis of children with septic shock from a one center similarly reported an association of HAT therapy with improved survival.⁸³ However, numerous subsequent clinical trials have failed to confirm a beneficial effect of HAT therapy and some studies have found evidence of harm.^84–86 As a result, despite biologic plausibility and clinical optimism, guidelines currently advice against routine use of metabolic resuscitation strategies until further evidence is available.^{33, 34}

Sepsis Sub-phenotypes and Precision Therapies

Most patients with sepsis exhibit common clinical features and a shared response to basic resuscitative interventions, but variability across pathogens, hosts, and host-pathogen interactions results in substantial heterogeneity in underlying pathology and response to therapy. Consequently, several sub-phenotypes have now been described in which some children with sepsis exhibit distinct clinical characteristics, an illness trajectory, or pathophysiologic features (Table 3).^{28, 87–89} Moreover, new mechanistic insights support the potential that novel precision therapeutics could improve outcomes in pediatric sepsis. For example, for children with sepsis-induced immune paralysis, treatment with the immune stimulant GM-CSF, reversed immune tolerance and reduced risk of hospital-acquired infections.⁹⁰ Children with sepsis-induced TAMOF resolved MODS faster if treatment included plasma exchange.⁹¹ In sepsis/macrophage activation syndrome (MAS) overlap syndromes, immune modulation with methylprednisolone, intravenous immunoglobulin, and/or human recombinant IL-1 receptor antagonist (anakinra) have reduced mortality and duration of organ dysfunction.^92–94 Based on these preliminary data, the Personalized Immunomodulation in Pediatric Sepsis-Induced MODS (PRECISE) clinical trial is testing the hypothesis that critically ill children with sepsis who have laboratory evidence for hyperinflammation/MAS or immune paralysis sub-phenotypes will improve with targeted immunomodulatory therapies that seek to reduce systemic inflammation or augment the immune response, respectively (www.clinicaltrials.gov/NCT05267821; www.clinicaltrials.gov/ NCT05266001).

Table 3:

Sub-phenotypes of Pediatric Sepsis

Phenotype Category	Example Phenotypes	Phenotype Characteristics
Routine clinical dataa	Normothermic	Normal body temperature for initial 72 hours (40% of patients, 11% mortality)
	Hyperthermic, slow resolver	Hyperpyrexia for >24 hours (35% of patients, 3% mortality)
	Hyperthermic, fast resolver	Hyperpyrexia that resolved within 24 hours (19% of patients, 9% mortality)
	Hypothermic	Initial hypothermia with return to near-normal body temperature within 24 hours (6% of patients, 17% mortality; lowest concentration of pro-inflammatory cytokines)
Organ dysfunction trajectoryb	Severe, persistent encephalopathy	Low GCS for up to 3 days (19% of patients, 11% mortality)
	Moderate, resolving hypoxia	Moderate respiratory dysfunction for 1–2 days, followed by improvement (35% of patients, 3% mortality)
	Severe, persistent hypoxia and shock	Severe shock and/or respiratory failure for 3 days (19% of patients, 17% mortality that was lower for patients treated with hydrocortisone)
	Moderate thrombocytopenia and shock	Shock and coagulopathy with or without hepatic dysfunction (23% of patients, 8% mortality)
OMIC profilingc	Endotype A	Patients classified as endotype A had decreased expression of genes related to adaptive immunity and glucocorticoid signaling within 24 hours of PICU admission and exhibited higher mortality than endotype B
	Endotype B
Pathophysiology/biologic mechanismd	TAMOF	Clinical findings of thrombocytopenia, disseminated microvascular thromboses, AKI, and evolving MODS due to low levels of a disintegrin and metalloproteinase with thrombospondin type 1 motif (ADAMTS-13) with resulting accumulation of ultra-large von Willebrand factor, widespread platelet activation, and thrombotic microangiopathy
	Immune paralysis	Clinical findings of immune tolerance, often defined as the inability of blood to produce high levels of TNFα when stimulated ex vivo with lipopolysaccharide, along with risk of hospital-acquired infections
	Hyperinflammatory/Hyperferrintinemic sepsis or HLH/MAS overlap	Clinical of findings of hyperferritinemia, hypofibrinogemia, hepatobiliary dysfunction, coagulopathy, and high levels of pro-inflammatory cytokines

^aAdopted from: Yehya N, Fitzgerald JC, Hayes K, Zhang D, Bush J, Koterba N, et al. Temperature Trajectory Sub-phenotypes and the Immuno-Inflammatory Response in Pediatric Sepsis. Shock. 2022;57(5):645–651.

^bAdopted from: Sanchez-Pinto LN, Bembea MM, Farris RW, Hartman ME, Odetola FO, Spaeder MC, et al. Patterns of Organ Dysfunction in Critically Ill Children Based on PODIUM Criteria. Pediatrics. 2022;149(1 Suppl 1):S103-S110.

^cAdopted from: Wong HR, Sweeney TE, Lindsell CJ. Simplification of a Septic Shock Endotyping Strategy for Clinical Application. Am J Respir Crit Care Med. 2017;195(2):263–265.

^dAdopted from: Carcillo JA, Berg RA, Wessel D, Pollack M, Meert K, Hall M, et al. A Multicenter Network Assessment of Three Inflammation Phenotypes in Pediatric Sepsis-Induced Multiple Organ Failure. Pediatr Crit Care Med. 2019;20(12):1137–1146.

Using an approach in which children with septic shock were both stratified by a blood protein signature associated with risk of mortality and assessed for differential whole blood gene expression, Wong and colleagues observed heterogeneity of treatment effect with hydrocortisone that, if confirmed, could drive precise use of corticosteroids in children with septic shock. Specifically, treatment with hydrocortisone was associated with reduced odds of a complicated course only in patients stratified as high-risk and preserved glucocorticoid-receptor gene expression. In contrast, treatment with hydrocortisone did not alter outcomes in either high- or low-risk patients with an endotype that included suppressed glucocorticoid-receptor gene expression.⁹⁵ These findings suggest that a favorable response to hydrocortisone in pediatric septic shock may be limited to high-risk patients biologically poised to respond to exogenous corticosteroids through preserved glucocorticoid-receptor gene expression. The potential to link prognostic enrichment (i.e., distinguishing high- versus low-risk of adverse outcomes) with predictive enrichment (i.e., distinguishing likelihood of response to a therapeutic intervention) to optimize glucocorticoid administration is currently being tested in the SHIPPS trial.

Conclusions

So long as humans and microbes continue to co-exist, it is unlikely that sepsis will ever be completely eradicated from our healthcare systems. However, an enhanced understanding of the pathophysiology driving the dysregulated host response to infection and subsequent failure to return to homeostasis will no doubt help us to minimize adverse outcomes from sepsis. Current guidelines focus on early recognition aided by increasingly sophisticated tools that identify patterns not easily recognized by clinicians and emphasize a set of initial management steps that have proven beneficial for most, even if imperfect for some. We are now moving toward an era in which precision therapeutics will offer new strategies to improve outcomes, especially for the subset of children with sepsis-induced MODS for whom antibiotics, fluid, vasoactive medications, and supportive care remain insufficient.

Abbreviations:

MODS

multiple organ dysfunction syndrome

SIRS

systemic inflammatory response syndrome

IPSCC

International Pediatric Sepsis Consensus Conference

PICU

pediatric intensive care unit

PODIUM

Pediatric Organ Dysfunction Information Update Mandate (PODIUM)

ATP

adenosine triphosphate

CRP

c-reactive protein

PCT

procalcitonin

AI

artificial intelligence

PRoMPT

BOLUS Pragmatic Pediatric Trial of Balanced versus Normal Saline Fluid in Sepsis

SHIPSS

Stress Hydrocortisone in Pediatric Septic Shock

HAT

hydrocortisone, ascorbic acid, and thiamine

TAMOF

thrombocytopenia-associated multiple organ failure

MAS

macrophage activation syndrome

PRECISE

Personalized Immunomodulation in Pediatric Sepsis-Induced MODS