A Critical Assessment of Time-to-Antibiotics Recommendations in Pediatric Sepsis

Abstract

The Pediatric Surviving Sepsis Campaign Guidelines recommend delivery of antibiotics within 1 hour for children with septic shock and, for those without shock but with sepsis-related organ dysfunction, as soon as feasible within 3 hours. In this review, we summarize the available adult and pediatric literature supporting these recommendations. We also explore the implications of implementing time-to-antibiotic goals at the point of antibiotic initiation in clinical practice, as well as the potential downstream impacts of these goals on antibiotic de-escalation.

Adult and pediatric data support the benefit of delivering antibiotics rapidly to children with sepsis, particularly with septic shock. Antibiotic timeliness goals have implications both at the point of antibiotic initiation and de-escalation, which are discussed.

INTRODUCTION

Sepsis guidelines, national and state-wide performance measures, and quality improvement efforts recommend rapid administration of broad-spectrum antibiotics within 1-3 hours to patients with suspected sepsis and septic shock [1, 2]. These time-to-antibiotics recommendations are based on adult literature supporting an association between delayed antibiotic administration and mortality, with a smaller number of pediatric studies reporting similar findings [3–19]. While these data compel action, the downstream impacts of antibiotic adverse events are also increasingly well documented, including acute kidney injury, Clostridioides difficile infection, selection for multidrug-resistant organisms, impact on the microbiome, and even risk of future sepsis [20–25]. Therefore, substantial controversy has emerged related to balancing the rapid delivery of antibiotics to patients who need them, while minimizing the avoidable harms of antibiotic overuse among patients who do not [26–28]. Herein, we provide a review of the published data related to optimal time-to-antibiotics in sepsis and explore the implications of these data for the clinical care of non-neonatal pediatric populations.

DEFINING SEPSIS

In 2005, the International Pediatric Sepsis Consensus Conference developed expert consensus-based criteria for pediatric sepsis, defining sepsis as systemic inflammatory response syndrome (SIRS) in response to a suspected or documented infection; severe sepsis as sepsis plus organ dysfunction of ≥2 systems; and septic shock as sepsis plus cardiovascular organ dysfunction [29]. The majority of studies providing the evidence base for international sepsis guidelines utilize cohorts meeting these SIRS-based definitions, though it should be noted that SIRS-based criteria lack sensitivity and specificity in identifying patients with sepsis [30, 31]. In 2016, the Sepsis-3 Taskforce conceptually re-defined sepsis in adults as infection-related, life-threatening organ dysfunction, and created data-driven criteria for diagnosing sepsis and septic shock using a large electronic health record database [32]. Using a similar conceptual framework and data-driven process, the Phoenix criteria for pediatric sepsis were published in 2024. The Phoenix criteria define pediatric sepsis as a suspected infection associated with a 2-point change in the Phoenix sepsis score, and septic shock as the subset of patients with sepsis who have at least 1 point in the cardiovascular dysfunction category (Table 1) [33]. The Phoenix sepsis criteria may be used to inform sepsis clinical care and quality improvement efforts—that is, to identify patients with sepsis at high risk of mortality who may benefit from sepsis-related therapies, including antibiotics. However, these criteria do not identify all children at risk for sepsis or those with sepsis-related organ dysfunction of organ systems not included in the Phoenix score, so are alone insufficient to identify children requiring treatment for sepsis at the time of initial presentation. More clinical definitions of sepsis, such as the one used by the Improving Pediatric Sepsis Outcomes (IPSO) collaborative, which include a combination of clinician actions such as antibiotic and blood culture orders, and at least 2 fluid boluses OR a vasopressor, may prove useful both to prompt consideration of sepsis “in the moment” when therapeutic decisions are occurring, and serve as operational definitions for sepsis for research purposes [18]. Overall, while there is likely overlap with those meeting SIRS-based criteria, the Phoenix criteria, and definitions of pediatric sepsis based on clinician actions, it is particularly relevant for the reader to be mindful of the populations studied in critically appraising the literature related to antibiotic timeliness in pediatric sepsis, taking particular note of the lack of published data related to antibiotic timing in children meeting the Phoenix criteria.

Table 1.

Phoenix Criteria for Pediatric Sepsis

Variable	0 Points	1 Point	2 Points	3 Points
Respiratory
0-3 points possible	PaO2:FiO2 ≥400 or SpO2:FiO2 > 292a	Any respiratory support a , b and: PaO2:FiO2 < 400 or SpO2:FiO2 < 292	IMV and: PaO2:FiO2 100-200 or SpO2:FiO2 148-220a	IMV and: PaO2:FiO2 < 100 or SpO2:FiO2 < 148a
Cardiovascular
0-6 points possible (including vasoactive, lactate, and mean arterial blood pressure measures)		1 point each (up to 3 total):	2 points each (up to 6 total):
	No vasoactive medicationsc	1 vasoactive	≥ 2 vasoactive medications
	Lactate < 5 mmol/L	Lactate 5-10.9 mmol/L	Lactate ≥ 11 mmol/L
Mean arterial blood pressure, mm Hgd
<1 month	>30	17-30	<17
1-11 months	>38	25-38	<25
1 to <2 years	>43	31-43	<31
2 to <5 years	>44	32-44	<32
5 to <12 years	>48	36-48	<36
12-17 years	>51	38-51	<38
Coagulation
0-2 points possible	Platelets ≥100 × 103/μL	Platelets <100 × 103/μL
	International normalized ratio ≤1.3	International normalized ratio >1.3
	D-dimer ≤2 mg/L FEU	D-dimer >2 mg/L FEU
	Fibrinogen ≥100 mg/dL	Fibrinogen <100 mg/dL
Neurological
0-2 points possible	Glasgow Coma Scale score >10; pupils reactivee	Glasgow Coma Scale score ≤105	Fixed pupils bilaterally
Sepsis	Suspected infection and Phoenix Sepsis Score ≥2 points
Septic shock	Sepsis with ≥1 cardiovascular point(s)

^aSpO₂:FiO₂ may only be calculated if the SpO₂ is ≤97%.

^bAny respiratory support includes supplemental oxygen, high-flow oxygen, non-invasive mechanical ventilation, or invasive mechanical ventilation.

^cVasoactive medications include epinephrine, norepinephrine dopamine, dobutamine, milrinone, and/or vasopressin (administered for shock).

^dAge-based criteria are based on chronologic age and not adjusted for prematurity. These criteria do not apply to birth hospitalizations, children whose postconceptional age is less than 37 weeks, or patients 18 years of age and older.

^eThe measure of neurologic dysfunction was derived from both sedated and non-sedated patients.

EVOLVING GUIDELINE RECOMMENDATIONS

Historically, the Surviving Sepsis Campaign (SSC) guideline recommended that antibiotics be administered to all patients within 1 hour of sepsis recognition based on observational data demonstrating that hourly delays in antibiotic administration were associated with increased odds of mortality, as well as associations between inadequate antibiotic treatment and poor outcomes [3–5, 34]. However, over the past decade, substantial evidence has provided a more nuanced understanding of the complex relationship between antibiotic timing and clinical outcomes. Most literature supports a greater benefit of rapid antibiotic delivery in patients with septic shock, with less certainty around the optimal threshold for antibiotic administration in less severely ill patients. While there is no reason to delay antibiotics once a bacterial infection is diagnosed or highly suspected, allowing time for clinicians to perform an initial diagnostic evaluation to determine whether antibiotics are indicated can balance the harms of unnecessary antibiotic exposure in uninfected patients with the benefits of early antibiotic therapy (Figure 1). Reflecting this body of evidence, both the 2021 adult and 2020 pediatric SSC guidelines recommend that antibiotics be administered within 1 hour for patients with septic shock and within 3 hours for patients with suspected sepsis without shock [1, 2].

Figure 1.

Conceptual framework of antibiotic timing goals. Adapted from: Evans L, Rhodes A, Alhazzani W, et al. Surviving Sepsis Campaign: International Guidelines for Management of Sepsis and Septic Shock 2021. Crit Care Med 2021; 49:e1063–e1143.

ANTIBIOTIC TIMING IN SEPSIS: ADULT LITERATURE

The evidence supporting these differential time-to-antibiotics goals for sepsis and septic shock is derived primarily from retrospective adult studies conducted in high-resource settings. Liu and colleagues performed a retrospective study of 35 000 adults with sepsis cared for in 21 emergency departments (EDs) between 2010 and 2013. Among patients with sepsis without shock, the adjusted odds ratio (aOR) for in-hospital mortality was 1.09 (95% confidence interval [CI], 1.00-1.19, P = .046) for each 1-hour delay in antibiotic administration over the first 6 hours, whereas in patients with septic shock, the mortality odds were substantially higher with each hour delay in antibiotic administration (aOR 1.14, 95% CI, 1.06-1.23, P = .001) [6]. Similar findings were demonstrated in a study of nearly 41 000 adults with sepsis managed between 2014 and 2016 under a New York state-mandated sepsis protocol, which required clinicians to obtain a blood culture before antibiotics, measure a lactate level, and administer antibiotics within 3 hours. In the analysis restricted to antibiotic timing, the odds of mortality increased every hour antibiotics were delayed, starting from the first hour (OR 1.04, 95% CI, 1.03-1.06, P < .001). However, and consistent with the previous study, this association was driven by patients requiring vasopressors (OR 1.07, 95% CI, 1.05-1.09), with no statistical association detected in those without shock (OR 1.01, 95% CI, 0.99-1.04) [7].

In the only patient-level randomized trial published to date, Alam and colleagues randomized adults with suspected sepsis based on SIRS criteria, to receive either pre-hospital antibiotics or usual care. Patients treated with pre-hospital antibiotics received antibiotics a median of 96 minutes earlier than those randomized to usual care, with usual care patients receiving antibiotics a median of 70 minutes after arrival in the ED. Overall mortality was 8% in both groups and there was no difference in any secondary outcome [8]. While the randomized design is a clear strength, generalizability to patients with septic shock is limited given that these accounted for only 4% of the cohort. Further, since antibiotics were generally delivered rapidly after ED arrival in the usual care group, we cannot conclude that longer delays would not have negatively impacted these less severely ill patients.

ANTIBIOTIC TIMING IN SEPSIS: PEDIATRIC LITERATURE

Single-Center, Retrospective Data

Weiss and colleagues compared the effect of hourly delays in antibiotic administration on pediatric intensive care unit (PICU) mortality in 130 children with sepsis-related organ dysfunction (21%) or septic shock (79%). While this study may have been limited by the small sample size compared to adult studies, hourly delays did not have a statistically significant impact on mortality until crossing the 3-hour threshold (aOR 4.84, 95% CI, 1.45-16.16 in the illness severity adjusted model) [14]. Han and colleagues expanded this work and assessed the impact of antibiotic timing on 1-year mortality in an overlapping cohort of 160 patients from the same center. In the primary analysis, there was no association between hourly delays in antibiotic administration and 1-year mortality, findings consistent with a smaller, single-center study of 79 children with sepsis where no association between antibiotic timing and either 1-year mortality or PICU length of stay was identified [15, 16]. In a larger single-center study of 441 Indian children with sepsis, delays in antibiotic delivery greater than 1 hour from sepsis recognition were associated with increased odds of mortality (aOR 1.83, 95% CI, 1.14-2.92). When stratified by illness severity, antibiotic timing was associated with mortality only in those with septic shock, though this difference may be explained by sample size limitations, as 77% of the cohort had septic shock (Table 2) [17].

Table 2.

Major Pediatric Studies Evaluating Time-to-Antibiotics in Sepsis

Author	Sample Size	Population	Exposure	Outcome	Results
Single center studies
Weiss et al  2014	130 Note: 79% had septic shock	ED and in patients with severe sepsis and septic shock requiring PICU admission	Antibiotic timing is measured in hours from the time of sepsis recognition	PICU mortality	Delay >3 hours: aOR 4.84 (95% CI, 1.45-16.16) in illness severity adjusted model Delay ≤3 hours: NS Overall mortality: 12%
Van Paridon et al 2015	79 Note: 56% had septic shock	Children admitted to the PICU with sepsis (73% admitted from an ED)	Antibiotic administration timing as measured from the time of presentation	PICU length of stay	No association between antibiotic timing and PICU length of stay Overall mortality: 6%
Han et al  2017	160 Note: 78% had septic shock	ED and in patients with severe sepsis and septic shock requiring PICU admission	Antibiotic timing is measured in hours from the time of sepsis recognition	One year mortality	No association between hourly delays and 1-year mortality Overall mortality: 24%
Sankar et al  2021	441 Note: 77% had septic shock	Children with sepsis, severe sepsis, or septic shock presenting to an ED in India	Administration of antibiotics within 1 hour of sepsis recognition	All-cause in-hospital mortality	aOR 1.83 (95% CI, 1.14-2.92) Overall mortality: 24%
Multicenter studies
Evans et al  2017	1179 Note: 69% had septic shock	Children with severe sepsis or septic shock presenting to the ED	Completion of a 1-hour bundle including antibiotic administration	All-cause in-hospital mortality	Bundle completion: aOR 0.59 (95% CI, 0.38-0.93, P = .02) Antibiotics only: aOR 0.78 (95% CI, 0.55-1.12) Overall mortality: 3%
Paul et al  2023	37 339 Note: 34% had “critical sepsis”	ED and inpatients with suspected sepsis or critical sepsis	Completion of a sepsis bundle including the use of a recognition tool, antibiotics within 180 minutes, and a fluid bolus within 60 minutes	30-day all-cause and sepsis-attributable mortality	Critical sepsis	Suspected sepsis
					Bundle compliant vs non-compliant: 2.4% vs 4.7% (P < .001)	Bundle compliant vs non-compliant: 0.4% vs 2.0% (P < .001)
					Antibiotic compliant vs non-compliant: 1.8% vs 2.0% (NS)	Antibiotic compliant vs non-compliant: 0.4% vs 0.7% (<.01)
Lane et al  2024	19 515	ED patients with suspected sepsis or critical sepsis	Defining an inflection point beyond which antibiotic timing is associated with mortality	3-day sepsis-attributable mortality	Time-to-antibiotics associated with increased 3-day mortality: 330 minutes 3-day mortality aOR for each 30-minute delay in antibiotic administration >330 minutes: 3.44 (OR 3.44, 95% CI, 1.20-9.93, P = .02)

Multicenter Data

Evans and colleagues used statewide data from New York collected following the implementation of the sepsis care mandate, which included collecting a blood culture and administering a fluid bolus and antibiotics within 1 hour of sepsis recognition. A total of 1179 children were included and nearly 70% had septic shock. Completion of the 1-hour bundle within both 1 hour and 3 hours was associated with a reduction in odds of in-hospital mortality (OR 0.59, 95% CI, 0.38-0.93, P = .02 and OR 0.64, 95% CI, 0.42-0.96, P = .03, respectively). There was no differential effect of bundle completion in patients with or without shock. When antibiotic administration within 1 hour was considered alone, compliance with this measure was not associated with reductions in mortality, suggesting the possibility that improved processes of care around all elements of sepsis management may be more impactful than a short antibiotic timing goal (OR 0.78, 95% CI, 0.55-1.12, P = .18) [13].

Finally, the largest pediatric studies published to date report findings from the multicenter IPSO collaborative, a quality improvement collaborative focused on improving sepsis outcomes through the implementation of a sepsis bundle in children with suspected sepsis and critical sepsis (an operational definition that approximates septic shock) in both inpatient and ED settings. Using data from 40 hospitals, Paul and colleagues demonstrated that among 24 518 children with suspected sepsis and 12 821 children with critical sepsis, both 3- and 30-day sepsis-attributable mortality (adjudicated by chart review by 2 independent parties at each site) were decreased in patients who received a fluid bolus within 60 minutes and antibiotics within 180 minutes. Applying a shorter timeline for fluid bolus and antibiotic administration (within 20 and 60 minutes, respectively) was not statistically associated with reductions in mortality in either group. When considering antibiotic administration within 180 minutes alone, compliance was associated with reductions in 3- or 30-day mortality in the suspected sepsis cohort only, with no difference detected in the critical sepsis cohort (Table 2) [18].

Building upon these findings and focusing on the impact of antibiotic timing specifically, Lane and colleagues examined the time point after which delays in antibiotic administration were associated with increased mortality in a cohort of 19 515 children presenting with suspected sepsis or critical sepsis to 51 EDs within the IPSO collaborative. Antibiotic timing was associated with increased 3-day sepsis-attributable mortality only when antibiotics were administered greater than 330 minutes after ED arrival (0.5% vs 1.2%), and beyond this time point, each 30-minute time interval was associated with a 3.4-times greater odds of death (OR 3.44, 95% CI, 1.20-9.93, P = .02). Further, patients who received antibiotics within 30 minutes of sepsis recognition had greater odds of mortality than those receiving antibiotics within 30-330 minutes (1.3% vs 0.4%, P < .01), highlighting the non-linear relationship between antibiotic timing and mortality. Finally, when limiting the analysis to those with confirmed bacteremia, delays in antibiotic administration beyond 90 minutes were associated with numerically higher mortality, though this association did not reach statistical significance [19].

Collectively, the limited available pediatric literature supports associations between antibiotic timing and clinical outcomes in children with sepsis, though a specific antibiotic timing threshold at which mortality increases is yet undefined. These findings are likely driven largely by the heterogeneity in pediatric sepsis cohorts with substantially different baseline risks of mortality based on comorbid conditions, timing of presentation, and illness severity; non-randomized designs with the attendant risks of residual confounding and bias; and lower short-term sepsis-attributable mortality relative to adult studies, which makes detecting a clinically meaningful difference in this outcome challenging [35]. Further, while adult studies more convincingly demonstrate the benefit of rapid antibiotic delivery in patients with septic shock, this association was not detected in the subgroup of patients with critical sepsis in the IPSO studies. This difference may reflect that IPSO “critical” sepsis includes patients who received either a vasopressor OR 3 fluid boluses, whereas most adult studies define septic shock based on the administration of a vasopressor, such that “critical” sepsis may therefore identify children with lesser sepsis severity than their adult counterparts. Further, the potential for inclusion of children with viral processes or non-infectious sepsis mimics in whom timely shock resuscitation but not antibiotic therapy would impact outcome may have diluted the potential impact of rapid antibiotics in the critical sepsis group. Lending support to this possibility, Lane and colleagues showed that the inflection point in the time at which antibiotic delays were associated with mortality was substantially shorter (though not statistically different) in children with confirmed bacteremia as compared to the overall sepsis cohort (90 minutes vs 330 minutes), whereas in the critical sepsis subgroup, no difference in inflection point was identified [19]. Finally, residual confounding may also contribute, as was observed in the Lane et al study, as presenting symptoms and illness severity can contribute to both the rapidity with which sepsis is recognized and antibiotic therapy is administered, factors that are challenging to capture and statistically adjust for, but nevertheless may contribute to clinical outcome [36].

DEFINING THE GOAL TIME FOR ANTIBIOTIC DELIVERY

Sepsis bundles and other quality improvement efforts are bound by imposing binary goals for antibiotic timing, dichotomizing a multidimensional set of circumstances reflecting the trade-offs between avoiding delays in necessary antibiotic treatment while allowing time to establish an infectious diagnosis to determine if antibiotic treatment is indeed necessary (Figure 1). This diagnostic uncertainty, in our view, underpins much of the controversy around antibiotic timing and is arguably the fundamental challenge in sepsis research and clinical care. Acknowledging this tension, we suggest that the urgency of antibiotic administration be calibrated to the patient’s illness severity as well as the certainty with which the patient requires antibiotics—either because of host factors or evidence of bacterial infection—rather than a “one size fits all” approach (Figure 1). More specifically, in children with septic shock, and in those whose need for antibiotics is certain (for example, febrile neutropenia, necrotizing fasciitis), broad-spectrum antibiotics should be administered without delay following collection of a blood culture and other appropriate cultures. In scenarios where there is some concern for sepsis, but uncertainty as to whether a bacterial infection is present, clinicians should be empowered through evidence-based guideline recommendations and local protocols to perform an initial diagnostic evaluation prior to ordering antibiotics. Current evidence would support a window of at least 3 hours is acceptable for this evaluation, and it remains to be seen how data from the IPSO cohort and other studies supporting an even longer time window in less severely ill patients are reflected in antibiotic timing goals in future guidelines [12, 14, 18, 19].

ANTIBIOTIC TIMELINESS CONSIDERATIONS AT THE POINT OF ANTIBIOTIC INITIATION

A key challenge is patient selection—that is, differentiating patients with sepsis from those presenting with uncomplicated febrile illnesses or other non-infectious mimics of sepsis—a challenge exacerbated by the lack of a rapid diagnostic test that reliably identifies bacterial infection at the point of initial presentation, rather than a day or more later after culture data is available [37]. Though SIRS-based criteria for defining pediatric sepsis have been replaced by more data-driven definitions under the Phoenix criteria, consider that over 90% of children with fever presenting to the ED meet vital sign-based criteria for SIRS and of these, over 80% are ultimately discharged home [31]. To date, the Phoenix criteria have not been evaluated in a “real world” setting as a tool for identifying children with sepsis warranting aggressive treatment, so it remains to be seen whether these definitions will translate to improved recognition, processes of care, and clinical outcomes. Further, though overt hypotension is readily apparent and the cause for immediate antibiotic delivery if thought to be due to infection, physical exam findings are otherwise notoriously insensitive for detecting milder organ dysfunction that may also herald sepsis [38]. Sepsis screening tools are recommended by national guidelines and may also serve to improve recognition, but there is no universally accepted screening tool available, owing in part to the non-specific signs of sepsis, the incredibly rare outcomes that are used to train these predictive models, and particularly in the inpatient setting, limited data deriving screening tools for children with baseline organ dysfunction [1, 39]. Finally, a potential future strategy uses machine learning tools to predict the risk of shock in children already diagnosed with suspected sepsis [40]. This approach may improve the balance between not missing opportunities for early recognition and treatment of potentially critically ill children while minimizing the risk of antibiotic overtreatment of patients who may not benefit, though requires additional study before being implemented in practice.

ANTIBIOTIC TIMELINESS CONSIDERATIONS: ANTIBIOTIC DE-ESCALATION OR DISCONTINUATION

Given the imprecision of sepsis diagnostic criteria and the lack of reliable, real-time diagnostic tests for many bacterial infections, it is not surprising that at least one-third of adults treated in the ED are ultimately determined to have no infection or only possible infections when evaluated post hoc [41, 42]. In addition to these patients who likely required no antibiotics at all (for example, those with viral infections that could explain a sepsis-like presentation or those with non-infectious mimics of sepsis including pancreatitis or myocarditis), far more are candidates for antibiotic de-escalation once a final diagnosis is made. While this process of de-escalation will always be a critical tool for optimal sepsis care, as clinicians can hardly be faulted for starting antibiotics amidst diagnostic uncertainty when they suspect a life-threatening infection, it is not a panacea for offsetting the potential downstream harms of antibiotic overtreatment. To start, clinicians are not always willing to stop antibiotics in a timely fashion, even when infection is unlikely. For example, among patients classified as having unlikely or no infections post hoc after being initiated on broad-spectrum antibiotics in the ED, the median duration of antibiotics was 5.5 days, duration longer than that required to treat many established infections, including community-acquired pneumonia, acute bacterial rhinosinusitis, and intra-abdominal infection following source control [41]. Second, when considering factors contributing to whether antibiotics are de-escalated, lower illness severity, and improving clinical trajectory are predictors of antibiotic de-escalation in the ICU. For patients presenting with suspected or likely sepsis, de-escalation may therefore be more difficult, highlighting the need to also focus on opportunities to optimize antibiotic use at the time of antibiotic initiation [43]. Finally, diagnostic uncertainty contributes to a lack of antibiotic de-escalation, which may be driven by incomplete evaluations prior to commencing antibiotics under pressure of antibiotic timing goals. For example, lack of bacterial cultures and lack of a clear diagnosis were modifiable risk factors associated with inappropriate antibiotic use in the ICU [44]. A wider time window for initial therapy in those without shock may provide additional time for the completion of a diagnostic evaluation, including obtaining appropriate cultures and ruling out non-infectious sepsis mimics.

CONCLUSION

Overall, existing literature demonstrates an association between delays in antibiotic administration and adverse clinical outcomes, though, in children, significant knowledge gaps exist regarding optimal strategies for identifying patients likely to benefit from early antibiotics and what the optimal time for antibiotic delivery should be across illness strata, infection type, and host factors. While better understanding the optimal timing for antibiotic delivery in children without shock is important, it is arguably more important to develop optimal tools to recognize children who truly have sepsis (vs other non-infectious mimics of sepsis), as “overtreatment” of uninfected children will be reduced with improved diagnostic tools, particularly those that are culture-independent and available at the point of care. While this ideal state remains elusive, data-driven definitions for pediatrics such as the Phoenix criteria represent an important step in the right direction, and future work should focus on understanding the impact of various therapies, including time-to-antibiotics, in children meeting these criteria. In addition, current efforts should also focus on ensuring necessary diagnostic testing is performed before antibiotic administration. Each of these represents key areas of multidisciplinary collaboration between infectious diseases, emergency medicine, and critical care clinicians and researchers.

Notes

Financial support. J.C.F. receives support from the National Institutes of Health (K23DK119463, P50DK114786).

Potential conflicts of interest. All authors: none.