Paediatric sepsis has been defined as proven or suspected infection in presence of systemic inflammatory response syndrome and septic shock as sepsis plus cardiovascular organ dysfunction.1 This long-standing definition from the 2005 International Pediatric Sepsis Consensus Conference (IPSCC) has been updated very recently, based on the new Phoenix criteria, but has not yet been translated into guidelines.2 3 25 million cases and 3.4 million deaths related to paediatric sepsis were reported in 2017 in a large international retrospective study where sepsis was identified according to medical diagnosis codes from death databases (‘sepsis’ or ‘infection’ and ‘organ failure’). Thus, sepsis remains the world’s leading cause of mortality in children under-5 years.4 Although fluid resuscitation was found to increase mortality in the only randomised clinical trial to date to assess it in children with severe febrile illness and impaired perfusion,5 it is one of the cornerstones of current paediatric septic shock management, in addition to early antimicrobial therapy.6 7 In 2020, the international guidelines of the Surviving Sepsis Campaign (SSC-2020) revised the management of septic shock and sepsis-associated organ dysfunction in children, with a particular attention to fluid resuscitation. In healthcare systems with availability of paediatric intensive care unit (PICU), fluid bolus of 10–20 mL/kg in case of abnormal perfusion or hypotension, up to 40–60 mL/kg over the first hour of septic shock recognition and preferably with balanced crystalloids were recommended.8 Despite the importance of these guidelines, to our knowledge, there is no study to date evaluating their application in paediatric emergency departments (PEDs).

We conducted the Fluid Resuscitation for Suspected Septic shock in Paediatric Emergency Departments (FRESSPED) Study to assess physician adherence to the SSC-2020 fluid resuscitation guidelines in children with suspected septic shock in French PEDs.

FRESSPED was a prospective observational study conducted in 21 French PEDs during 5 sequential weeks between November 2021 and March 2022. The methodology was consistent with ethical standards, and this report follows the guidelines of the Strengthening the Reporting of Observational studies in Epidemiology statement.9 A registration on ClinicalTrials.gov database was performed prior to inclusion (NCT05066464).

All children (<18 years) with suspected septic shock who received fluid resuscitation (at least 5 mL/kg within 1 hour) in participating PEDs during the 5-week inclusion period were eligible. Septic shock suspicion was defined according to the prescription of antimicrobial therapy within 72 hours and the presence of at least three of the eight criteria of the American Academy of Pediatrics (AAP) septic shock trigger tool (temperature abnormality, hypotension, tachycardia, tachypnoea, capillary refill abnormality, mental status abnormality, pulse abnormality, skin abnormality). Only two of these criteria were required in case of a high-risk condition (malignancy, asplenia including sickle cell disease, bone marrow or solid organ transplant, central or indwelling line/catheter, severe mental retardation or cerebral palsy, immunodeficiency or immunosuppression).10 Exclusion criteria were preterm infants under 39 weeks’ corrected age at screening time and children whose parents (or legal guardians) have expressed their objection to data collection.

Children with ‘severe sepsis’ and ‘septic shock’ were in retrospect defined as meeting the 2005 IPSCC criteria.1 Septic shock recognition was defined as the moment when fluid resuscitation was prescribed at the end of the clinical examination that led to it.

Primary outcome was SSC-2020 fluid resuscitation guidelines adherence (low 0–24%; moderate 25–74%; high 75–100%) according to: (1) volume of 10–20 mL/kg for each bolus; (2) exclusive administration of balanced crystalloids at 1 hour and 24 hours of management; and (3) initiation of fluid resuscitation within 1 hour of septic shock recognition. Primary outcome was assessed independently for each of its three individual components and for each bolus (including within the same patient). SSC-2020 fluid resuscitation guidelines for healthcare systems with availability of intensive care were used as reference.

Secondary outcome included: other key elements of fluid resuscitation (time to fluid resuscitation, volume of each bolus, prescribed and real bolus duration, number of bolus received at 1 hour, and total fluid bolus volume received at 1 hour and 24 hours); non-fluid management (time to clinical examination, time to antimicrobial therapy start, blood culture collection, oxygen therapy, lactate measurement, and peripheral venous and intraosseous accesses); and barriers to SSC-2020 fluid resuscitation guidelines adherence reported by physicians.

An invitation to participate in the study was sent by email to 26 French PEDs’ heads. FRESSPED was carried out in three successive stages. First, usual fluid resuscitation practices were assessed through the completion of paper case report forms (less than 5 min), within 72 hours of inclusion, by physicians of the participating centres caring for an eligible patient during the 5-week inclusion period. Second, patients were followed up during 1 month after inclusion (extending the study period to 28 April 2022) to investigate their outcomes by data collected in routine care (patient chart) by the principal investigator. Third, from May to June 2022, participating physicians responded to a short, anonymised online survey, blinded of centre outcomes, to explore the barriers to SSC-2020 fluid resuscitation guidelines adherence as well as their attitudes and knowledge.

Paper case report forms were scanned and deposed on a secure online platform (HCP Anywhere) by local coordinating investigators. Then, data were collected, anonymised and entered in an electronic database (Microsoft Excel; Microsoft Corp) by the principal investigator. Data were analysed using descriptive statistics to evaluate the study population. Each score was calculated using the worst values observed during the period of interest. Qualitative variables are reported as numbers and percentages. Quantitative variables are reported as median with IQR and Agostino-Pearson test was used to assess distribution. Statistical analyses were performed using GraphPad Prism V.9.4.0 (GraphPad Software, California, USA).

We sought to assess physician adherence to SSC-2020 fluid resuscitation guidelines in children with suspected septic shock in PEDs. Such adherence was high for fluid resuscitation initiation and moderate for bolus volume and fluid choice. Main barriers reported were difficult intravenous access, lack of team training, workload constraints and absence or out-of-date protocols. Despite these discrepancies, no patient died among the few cases of severe sepsis.

The implementation of guidelines is an essential part of quality improvement process. One of the challenges facing emergency departments is whether guidelines should be implemented when a condition is suspected but not proven.11 Thus, this work focused on children with suspected septic shock, regardless of whether they had it. This population was selected because the problem of concern was the initial management in PEDs, where the clinician is not certain of septic shock. This initial management has a major prognostic impact, as recently highlighted in 259 French children.12 In addition, given the short inclusion period of the study (5 weeks), this population of interest allowed the inclusion of more patients (n=63) than only severe sepsis cases (n=10).

Characteristics of patients included in the study differ from the literature, with a median age of 7.9 months (2.8–35.9), as well as a fewer rate of comorbid conditions (30%) and undocumented infections (23%).13 One explanation for this may be the winter inclusion period, where influenza outbreaks (22%) occurred in mostly healthy infants who frequently presented poorly tolerated fever and signs of abnormal perfusion but no septic shock. In addition, it is striking that only two children had positive blood cultures. Moreover, it is striking that bacteria were isolated from blood cultures in only two patients. In addition, imputability of viruses, identified in 44% of cases, is difficult to establish given their prevalence in healthy children and their frequent association with bacteria that are themselves responsible for septic shock.14 In addition, the included patients were not critically ill, with a total fluid resuscitation volume of 19.7 mL/kg (10.4–28.0) and a peak serum lactate of 2.8 mmol/L (2.0–4.0), as well as a median length of hospital stay of 5 (3–7) days.

The FRESSPED Study is, to our knowledge, the first prospective multicentre study evaluating adherence to the SSC-2020 fluid resuscitation guidelines for suspected septic shock in PEDs. The 21 participating centres represent 24% of French PEDs of which 17% of academic medical centres in France. First, this adherence could be improved by implementing SSC-2020-based care protocols, especially as only 52% of physicians were aware of these guidelines in our study. Indeed, such protocols, including ‘bundles’ completion, have been associated with shorter time to recognition of septic shock, initiation of fluid resuscitation and antimicrobial therapy, as well as reduced organ dysfunction, length of hospital stay and mortality.15–17 Second, the creation of a teaching programme for health professionals would be of interest, as it has proven effective in increasing compliance with guidelines, especially as the lack of team training was reported by 29% of physicians.18 Time to antimicrobial therapy from septic shock recognition which was 73 min (40–145) in our study was not consistent with the 1-hour threshold of the SSC-2020 guidelines. Some authors have reported higher mortality with delayed antibiotic therapy beyond 1 hour in this setting19 and therefore such a teaching programme should emphasise early initiation of antimicrobial therapy. Finally, training in the placement of intraosseous catheters would reduce the difficulties of intravascular access that were reported in 43% of cases. Only 3% of children included benefited from an intraosseous line, which is yet recommended in case of peripheral venous cannulation failure and has shown a decrease in insertion time, length of hospital stay and mortality in relation to the latter.20 This can be explained by the low severity of the patients and by the large proportion (38%) of responding physicians who considered themselves not competent to perform it.

The FRESSPED Study has several limitations. First, the SSC-2020 fluid resuscitation guidelines specifically apply to children with septic shock, as defined in the 2005 IPSCC,1 and other sepsis-associated acute organ dysfunction.8 In retrospect, most of the patients included did not fulfil these criteria, as only 10 (16%) and 2 (3%) of whom met the definition of severe sepsis and septic shock, respectively. Although inevitable if one is identifying children at risk of septic shock at an early stage of the condition, this major discrepancy between the population addressed by the SSC-2020 guidelines and that included in our study highlights the inaccuracy of the suspicion of septic shock according to the prescription of antimicrobial therapy within 72 hours and the AAP septic shock trigger tool.10 We can thus assume that the pressure to identify sepsis early in the illness led to fluid resuscitation in children who had not cardiovascular dysfunction requiring it. This is a serious issue since such an intervention can be harmful to children, as shown by the FEAST Study, even though the patients included in our study were different from those in the latter.5 Furthermore, this may explain the relatively low rate of adherence to the SSC-2020 guidelines, which may have been skewed by the low probability of septic shock in the least severe children, leading some physicians to deliberately not follow these recommendations. This finding is consistent with an observational study by Nijman et al in which, among children at risk of sepsis at PEDs, only 0.6% of whom developed a serious bacterial infection or were admitted to a PICU and only 0.2% complied with the 1-hour sepsis bundle of care. The authors conclude that early escalation of care for children with suspected sepsis at the PEDs could have adverse effects on the overall management.21 It is therefore necessary to better define the criteria for suspicion of septic shock in PEDs. Second, in the absence of consultation of local registries, some patients meeting the inclusion criteria may not have been included, although this risk was largely reduced by: (1) the short sequential inclusion period; (2) the multiple reminders sent to local investigators preceding the weeks of data collection; (3) a monthly newsletter; (4) the systematic check of all patients who received a fluid resuscitation at the end of each week of inclusions by the local investigator. Third, as physicians were not blinded, there is an inherent bias related to practice observation, limited by the unawareness of the SSC-2020 guidelines reported by half of them.22 Finally, although there were no missing data on the primary outcome, the small sample of the study (n=63) did not allow for comparative statistical analyses, particularly concerning the children outcomes according to compliance with guidelines.