The Current and Future State of Pediatric Sepsis Definitions: An International Survey

The study was approved by the human research ethics committee, including implicit consent from participants filling out the survey (Children`s Health Queensland, Australia; LNR/19/QCHQ/559907).

A working group was formed with subject matter experts from the Pediatric Sepsis Definition Taskforce, which drafted, piloted, and revised the survey iteratively among its members (Supplemental Information).²¹  Once finalized, the 83-item survey was translated into Spanish, Portuguese, Mandarin, and French. The survey data were collected using Research Electronic Data Capture hosted at The University of Queensland. The World Federation of Pediatric Intensive and Critical Care Societies (WFPICCS) led the electronic dissemination of the survey to a total of 27 national or regional societies (Supplemental Information) with a focus on (pediatric) intensive care, pediatrics, emergency medicine, and infectious diseases. Most members have multiple memberships across national and regional societies and WFPICCS. We estimated the response rate for medical staff working in intensive care as 30.3% based on WFPICCS, and 13.9% based on non-WFPICCS societies. The survey instrument is available in the Supplemental Information.

Respondents were asked to provide information regarding their discipline, work environment, experience, and the availability of diagnostic and therapeutic tools. The second part of the survey assessed how clinicians approach a child being evaluated for sepsis and how they categorize the likelihood of sepsis or septic shock in relation to specific diagnostic or management criteria. Next, respondents were asked to state the perceived usefulness of current definitions of sepsis, including the IPSCC definition,⁸  the adult Sepsis-3 definitions,¹¹  and the WHO definition,²²  for the purposes of sepsis recognition, correct disease classification, prognostication, quality control and benchmarking, epidemiologic research, and for enrollment into clinical trials. Finally, respondents were asked to provide their opinion on how a future definition of the term sepsis should be used and to whom it should apply. Five-point Likert scales ranging from 1 = “strongly agree” to 5 = “strongly disagree” were used to assess respondents’ preferences. The survey was open from December 1, 2019 until March 15, 2020.

The reporting of this manuscript follows the American Association for Public Opinion Research (AAPOR) guidelines.²³  Data are presented as counts and percentages for categorical variables. We compared the partial and complete respondents’ characteristics using χ² tests (Supplemental Table 3). We excluded surveys with 50% or fewer questions answered and surveys that did not contain key respondent characteristics. We categorized responses into 3 different income country subgroups: (1) low-income (LIC) combined with low-middle-income (LMIC), (2) upper-middle-income (UMIC), and (3) high-income (HIC) countries according to the World Bank classification of 2019 to 2020.²⁴  The Likert scale responses were ranked according to the proportion of positive responses, where positive responses included both strongly agree and agree.

Subgroup analyses were conducted to investigate if the responses differed between income country subgroups, the clinical expertise, and the clinical setting (Supplemental Information and Supplemental Figs 6–17 and Supplemental Tables 3–8). Analyses were performed with the R statistical package version 4.0.2 (R Foundation for Statistical Computing, Vienna, Austria).²⁵ 

A total of 2891 responses were received. Surveys from 56 (1.9%) respondents who answered <50% of the questions or who did not complete the respondent characteristics section were excluded. A total of 2835 responses were analyzed (Table 1), including 1645 (56.9%) with complete surveys (Supplemental Table 3).

The most common region of origin was South America, followed by Asia (Fig 1). The respondents’ geo-economic classified group was LIC or LMIC in 14.1% (N = 399), UMIC in 48.0% (N = 1362), and HIC in 37.9% (N = 1074). The most common language used was English (1252, 44.2%), followed by Spanish (953, 33.6%), Mandarin (377, 13.3%), Portuguese (183, 6.5%), and French (70, 2.4%). The most common occupation among respondents was physician (2628, 92.7%), and 97% of all respondents served in clinical roles. Fifty-seven percent worked in a PICU (N = 1620), 15% in an emergency department (ED) (N = 424), 28% in ward settings (N = 791), and 6.9% (197) cared for adult patients too.

The most commonly available diagnostic modalities (>85% for LIC/LMIC, UMIC, and HIC) were simple radiologic imaging, complete blood count with differential, blood gas analysis, and noninvasive vital sign monitoring (Fig 2). The least available diagnostic modalities were rapid or polymerase chain reaction-based viral testing and echocardiography (<70% for LIC/LMIC and UMIC, <90% for HIC). The most commonly available management tools (>90% for LIC/LMIC, UMIC, and HIC) included fluid boluses, peripheral intravenous cannulation, and urinary catheterization. Extracorporeal membrane oxygenation (ECMO) and invasive hemodynamic monitoring were least commonly available (<25% for LIC/LMIC and UMIC, <65% for HIC). Availability of resources varied substantially based on income setting and clinical environment (Supplemental Table 4). For example, measurement of serum lactate levels was unavailable to 14% of respondents (25% in LIC/LMIC, 20% in UMIC, and 3% in HIC). The use of central venous access and the services of a pediatric intensivist were unavailable to 18% of respondents.

Consistent with the IPSCC definitions, abnormal vital signs, laboratory evidence of inflammation, and microbiological confirmation were the criteria most commonly used for the diagnosis of children with sepsis (Fig 3A, and Supplemental Table 9). Parental concern and clinicians “gut feel” emerged as least important variables. Evidence of (and treatments for) infection-induced organ dysfunction were most commonly used for the diagnosis of pediatric septic shock, particularly interventions intended to provide hemodynamic support such as fluids or inotropes (Fig 3B, and Supplemental Table 10).

We assessed respondents’ opinions about the perceived usefulness of the IPSCC, Sepsis-3, and WHO definitions in relation to clinical recognition, correct disease classification, outcome prognostication, quality control and benchmarking, understanding of sepsis epidemiology, and enrollment into clinical research (Fig 4). None of the available sepsis definitions was perceived as useful (agree or strongly agree) by more than 50% of the responders across all these domains. The IPSCC definition was perceived to be the most useful for sepsis recognition. The Sepsis-3 definition was classified as slightly more useful for benchmarking, disease classification, research, and prognostication. The WHO definition was perceived to be the least useful in most domains, except for recognition. Overall, respondents identified strong agreement that the next version of pediatric sepsis definitions should address all of these domains, with the recognition of sepsis receiving the greatest agreement (Fig 4).

A majority (71%) of respondents agreed that a future revision of the concept sepsis should not be applied to children without organ dysfunction in hospital settings, indicating support for the concept of sepsis as infection in the presence of organ dysfunction (Fig 5). The highest agreement (72%) was reached for sepsis as life-threatening organ dysfunction that is remote from the primary site of infection (eg, new renal or cardiovascular dysfunction in a child with pneumonia), followed by sepsis as life-threatening infection with fluid-refractory cardiovascular dysfunction (62%, eg, septic shock), and sepsis as life-threatening organ dysfunction that is limited to the primary site of infection (60%, eg, new respiratory failure in a child with pneumonia).

This global survey explored how health care workers involved in the care of children with sepsis view diagnostic concepts of sepsis. Previous studies revealed inconsistent application of criteria for sepsis,¹⁹  the lack of specificity of the criteria,¹⁶  and obstacles toward the global application of such criteria.^6,17,18  This global survey articulates the challenges in applying sepsis definitions to children across diverse settings characterized by marked variations in resources. The notion of sepsis as infection with associated organ dysfunction¹⁵  was widely supported by respondents. The results indicate a strong preference of end-users to have access to criteria that enables improved recognition of sepsis, consistent with the needs of quality improvement initiatives. The findings will feed into the Delphi round of the Pediatric Sepsis Definition Taskforce and assist in shaping a data-driven approach to develop revised criteria for sepsis in children, which are adaptable to global settings.

The quest for sepsis criteria in the past has oscillated between the need to phenotypically characterize a nonspecific clinical syndrome and the requirement of clinicians, researchers, and health care organizations to operationalize a disease for the purposes of reproducibly classifying and measuring, or conducting clinical trials.²⁶  At one end of the spectrum, sepsis was defined in 2001 as “documented or suspected [infection] and some of the following…” with 24 individual parameters listed.⁷  In 2005, IPSCC emphasized the importance of early, sensitive recognition.^27–29  Given that most febrile children manifest signs of SIRS even in the absence of organ dysfunction, by necessity the IPSCC definition is sensitive at the expense of specificity. In contrast, the adult Sepsis-3 criteria meet expectations for Specific, Measurable, Achievable, Realistic, and Timely concepts, and were designed to facilitate specific identification of patients with substantially increased mortality. Yet, these criteria omitted pediatric age groups and have proven to be only partially applicable to LIC and MIC settings. In addition, fundamental controversy surrounds their value in relation to initiatives such as the Improving Pediatric Sepsis Outcomes collaborative, which fosters systematic quality improvement in sepsis recognition and treatment across over 57 US hospitals.^17,30,31 

In the present survey, the differences in the level of agreement across individual criteria were relatively minor. For example, over 80% of all respondents stated that hemodynamic support (through fluid bolus or vasoactive-inotropic therapy), laboratory or therapeutic variables indicating organ dysfunction, signs of poor perfusion (such as poor skin perfusion, altered neurologic function, low urinary output, or increased lactate), and sepsis scores identified children with septic shock. These findings likely reflect the daily situation when clinicians are confronted with a child who “looks septic.” Rather than applying specific criteria, respondents seem to base such a diagnosis on a wealth of variables associated with the use of clinical reasoning skills, even if the importance of “gut feeling” was ranked low in our survey. Angus referred to this phenomenon in relation to sepsis as an inherently “fuzzy” syndrome.³²  As objective diagnostic criteria are needed for operational, quality, and research purposes, a data-driven approach may yield more relevant diagnostic criteria than an a priori opinion-based approach.^33,34 

As per the respondents of this survey, current approaches to define sepsis in children fall short of the need to provide both improved recognition for the purpose of allowing earlier treatment, as well as accurate case classification for the purpose of benchmarking, research, prognostication, and epidemiologic assessment. Accordingly, the development of new pediatric sepsis criteria may necessitate a tiered approach, starting with the derivation and validation of criteria characterizing groups of infected patients at substantially increased risk of mortality, eg, children with established (multi-) organ dysfunction, followed by the development of tools facilitating early recognition of such phenotypes. There was wide agreement among respondents that organ dysfunction is required to distinguish sepsis from infection. While sepsis-related organ dysfunction may be confined to the site of primary infection only, the concept of systemic processes (such as dysregulated host response to infection) leading to organ dysfunction remote from the site of primary infection was considered a slightly stronger indication of the presence of sepsis.

Most children and neonates developing and succumbing from sepsis globally live in less resourced areas in South America, Asia, or Africa.^1,35,36  Yet, only a minor proportion of pediatric sepsis trials are conducted in these settings^13,14,37  and neither the expert panel for the IPSCC nor the Sepsis-3 Taskforce included experts from LIC, LMIC, or UMIC settings. Furthermore, as indicated by our survey, the availability of diagnostic and therapeutic modalities varies not only by country income level, but by discipline and clinical environment as well. Any initiative to revise pediatric sepsis definitions should aim to explore criteria that can be applied in or adapted toward differently resourced health care settings. The bedrock of such an undertaking is a firm understanding on how sepsis in children is diagnosed based on resource availability. For example, evidence supports the value of serum lactate concentration in identifying children with sepsis at higher risk of mortality.^38–41  However, the restricted availability of lactate measures outside HIC identified in our survey contrasts with IPSCC (and Sepsis-3) criteria for septic shock that incorporate lactataemia.^8,11  We posit that the lack of available tools, such as lactate, is likely underestimated in our survey given that the predominate responses were from UMIC and HIC settings, supporting the need to improve diagnostic resources.⁴² 

The response pattern of this survey with a predominance of UMIC and HIC, compared with 14.1% of responses originating in LIC/LMIC mirrors the current inequities in global access to intensive care resources.^43–45  Our findings are supported by a number of recent studies: an international survey across 350 facilities in 4 LIC/LMIC countries identified that only 13% had sufficient resources to manage childhood pneumonia with basic measures, such as antibiotics, saturation monitoring, and oxygen,⁴⁶  confirming that organ support in LICs remains largely unavailable. In a recent Brazilian study, the PICU bed-to-population ratio varied from 1 per 604 to 1 per 6812, and 41% of national PICUs were classified as well resourced, in contrast to 27% with intermediate, and 32% with low resourcing,⁴⁷  demonstrating major inequities even within the same country.⁴⁷  A global survey on adult and pediatric critical care delivery including 328 respondents (11.7% from LICs) observed significantly lower availability of laboratory and imaging diagnostics, invasive monitoring, blood products, and advanced ventilatory support in LMIC settings, affecting the capability to deliver a basic sepsis bundle.⁴⁸  Failure to meet WHO standards of hospital care resources has prompted a recent White Paper on Critical Care Services in Low Resource Settings.⁴²  Thus, the present survey likely reflects the reality of current provision of critical care support for children in LIC and LMICs where resources are poor and our clinical colleagues are overburdened.⁴⁹ 

This study has several limitations. First, due to the dissemination through several societies and the lack of data on multiple memberships, the response rate needs to be considered approximate. We acknowledge that this represents a major limitation to external validation as the survey was biased toward respondents working in critical care. Other disciplines, such as emergency medicine, and professions, such as nurses, which play an important role in the early diagnosis of children with sepsis, were underrepresented in this survey. Overall, responses were skewed with underrepresentation of community-based settings in LIC and LMIC, especially from African countries, where large numbers of sick children present in the absence of intensive care resources.⁵⁰  The survey was translated into 5 of the most commonly used languages, which represent about half of the world’s population, but we acknowledge that the lack of translations into additional languages such as Arabic may have contributed to selection bias. Second, inherent to any survey on self-reported practice, we are unable to comment on discrepancies between stated versus observed practice. Third, we consciously refrained from testing knowledge or familiarity with the 3 sepsis definitions, which were provided explicitly on the survey portal, and hence the survey does not allow us to interpret the degree of uptake of the definitions in daily practice. Fourth, the survey explicitly referred to “pediatric” sepsis, and to “children,” and did not assess respondents’ perception on the applicability of these criteria for neonatal age groups.^51,52  Finally, although the survey dissemination overlapped with the latest pediatric Surviving Sepsis Campaign release, it seems unlikely that this affected results relevantly, given that the SSC stated that IPSCC remain the only available pediatric sepsis criteria.

In conclusion, this international survey highlights the challenges in applying sepsis criteria to children depending on the available resources of health care settings, clinical environment, and disciplines. This indicates a need for organ dysfunction criteria which can be pragmatically implemented in emergency, ward, or other general pediatric settings around the world. Future diagnostic algorithms should carefully balance pragmatism versus simplification. The survey identified a need for end-users to have access to sepsis tools which are sensitive as well as titratable to the locally available resources.

We thank the World Federation of Pediatric Intensive and Critical Care Societies, the Society of Critical Care Medicine (SCCM), and the European Society of Pediatric and Neonatal Intensive Care (ESPNIC) for their support in facilitating this project, Lori Harmon from the SCCM for her help with the Pediatric Sepsis Definitions Task Force, and Trang Pham, data scientist at the Children’s Health Research Centre at the University of Queensland, for her valuable help in creating the RedCap form for this survey.

On behalf of the Pediatric Sepsis Definition Taskforce: Samuel Akech, KEMRI-Wellcome Trust Research Program (Nairobi, Kenya); Elizabeth R. Alpern, Ann & Robert H. Lurie Children’s Hospital, Department of Pediatrics, Northwestern University Feinberg School of Medicine (Chicago, IL); Fran Balamuth, Chil’ren's Hospital of Philadelphia (Philadelphia, PA); Tellen D. Bennett, University of Colorado (Denver, CO); Paolo Biban, Verona University Hospital (Verona, Italy); Joseph A Carcillo, MD Professor of Critical Care Medicine and Pediatrics, UPMC Children's Hospital of Pittsburgh (Pittsburgh, PA); Enitan Carrol, University of Liverpool (Liverpool, UK); Kathleen Chiotos, Children’s Hospital of Philadelphia (Philadelphia, PA); Mohammod Jobayer Chisti, International Centre for Diarrhoeal Disease Research (ICDDRB) (Dhaka, Bangladesh); Idris Evans, UPMC Children's Hospital of Pittsburgh (Pittsburgh, PA); David Inwald, Addenbrooke's Hospital, Cambridge University Hospital NHS Trust (Cambridge, UK); Paul Ishimine, University of California San Diego (San Diego, CA); Michael Levin, Imperial College London (London, UK); Rakesh Lodha, All India Institute of Medical Sciences (New Delhi, India); Kathryn Maitland, KEMRI-Wellcome Trust Research Programme (Kilifi, Kenya); Kusum Menon, Department of Pediatrics, Children’s Hospital of Eastern Ontario, University of Ottawa (Ottawa, Canada); Simon Nadel, St. Mary’s Hospital (London, UK); Claudio Flauzino Oliveira, Associação de Medicina Intensiva Brasileira (São Paulo, Brazil); Mark J Peters, Great Ormond Street Hospital and UCL Great Ormond Street Institute of Child Health, NIHR Biomedical Research Centre (London, UK); Adrienne G. Randolph, Boston Children's Hospital (Boston, MA); L. Nelson Sanchez-Pinto, Ann & Robert H. Lurie Children's Hospital of Chicago (Chicago, IL); Halden F. Scott, Children's Hospital Colorado (Aurora, CO); Juliane Bubeck Wardenburg, Department of Pediatrics Washington School of Medicine St. Louis (St. Louis, MN); Scott L. Weiss, Children's Hospital of Philadelphia (Philadelphia, PA); Wilson Milton Were, Department of Maternal, Newborn, Child and Adolescent Health, World Health Organization (Geneva, Switzerland); Matthew O. Wiens, University of British Columbia (Vancouver, Canada), Mbarara University of Science and Technology (Mbarara Uganda); James L. Wynn, University of Florida (Gainesville, FL).

AAPOR

American Association for Public Opinion Research

ECMO

extracorporeal membrane oxygenation

ED

emergency department

ESICM

European Society of Intensive Care Medicine

ESPNIC

European Society of Pediatric and Neonatal Intensive Care

HIC

high income country

IPSCC

International Pediatric Sepsis Consensus Conference

IQR

interquartile range

LIC

low-income country

LMIC

lower-middle income country

SCCM

Society of Critical Care Medicine

SMART

Specific, Measurable, Achievable, Realistic and Timely

SOFA

Sequential Organ Failure Assessment

UMIC

upper-middle income country

WFPICCS

World Federation of Pediatric Intensive and Critical Care Societies

WHO

World Health Organization