The Effect of Rapid Initiation Versus Delayed Initiation of Antibiotics in Pediatric Patients With Sepsis

Abstract

OBJECTIVE

The purpose of this study was to determine if administration of antibiotics within 1 hour of meeting sepsis criteria improved patient outcomes versus antibiotics administered greater than 1 hour after meeting sepsis criteria in pediatric patients. The Surviving Sepsis Campaign's international guidelines recommend appropriate antimicrobial therapy be administered within 1 hour of recognition of severe sepsis or septic shock. Data regarding outcomes in pediatric patients with sepsis regarding antibiotic timing are currently limited.

METHODS

This was a retrospective chart review of 69 pediatric patients admitted between July 1, 2013, and June 30, 2016, with a diagnosis of sepsis.

RESULTS

The primary outcome of in-hospital mortality was 7.1% in the within 1 hour group versus 14.6% in the greater than 1 hour group (p = 0.3399). Median hospital length of stay was significantly shorter in the within 1 hour group (15.4 versus 39.2 days, p = 0.0022). Median intensive care unit length of stay was also significantly shorter in the within 1 hour group (3.1 versus 33.6 days, p = 0.0191). There were no differences between groups for pediatric intensive care unit admission, end organ dysfunction, time to intubation, or time on the ventilator.

CONCLUSIONS

Pediatric patients who receive antimicrobial therapy within 1 hour of meeting sepsis criteria had improved hospital and intensive care unit length of stay. This study supports the Surviving Sepsis Guidelines recommendation to administer antibiotics within 1 hour in pediatric patients with sepsis or septic shock.

Introduction

Sepsis remains a major cause of morbidity and mortality in pediatrics.¹ The Surviving Sepsis Guidelines define sepsis as the presence of life-threatening organ dysfunction caused by a dysregulated host response in the setting of infection.¹ The use of systemic inflammatory response syndrome (SIRS) criteria, including body temperature, heart rate, respiratory rate, and white blood cell count, are commonly used in clinical practice to assist with the diagnosis of sepsis.² In children, SIRS criteria are based on patient age, and a diagnosis of SIRS is based on meeting 2 or more of these criteria.² Although SIRS is commonly described as a nonspecific marker of an inflammatory process that can occur after trauma, burns, infections, and other diseases, the most common reason for a positive SIRS screen in pediatric patients is infection, making SIRS an appropriate surrogate marker for sepsis.²

The Surviving Sepsis guidelines¹ recommend administration of effective intravenous antibiotic agents within the first hour following recognition of sepsis and septic shock for both adult and pediatric patients. Per these guidelines, effective antibiotic therapy is defined as the use of intravenous antimicrobial agents with activity against all likely pathogens, including bacterial, fungal, and viral ones.¹ These agents must penetrate in adequate concentrations into the tissues of the presumed source.¹

In adults, delayed antimicrobial administration, greater than 1 hour, after diagnosis of sepsis or septic shock has been repeatedly associated with increased mortality.^3–7 In children, few studies have evaluated the impact of antimicrobial delay on clinical outcomes.^8–15 In addition, although some of these pediatric studies have found benefit in prompt antibiotic administration on outcomes such as mortality and end organ dysfunction, others have found conflicting results.^8–15 With the currently available literature, the effects of a delay in administration of antimicrobials greater than 1 hour in pediatric patients with sepsis or septic shock remain unclear. The purpose of this study was to determine if administration of antibiotics within 1 hour of meeting SIRS criteria improves patient outcomes versus antibiotics administered greater than 1 hour after meeting sepsis criteria in pediatric patients.

Materials and Methods

Study Design. This Institutional Review Board–approved study was a retrospective review of the electronic health record of all pediatric patients ages <18 years admitted to a single-center children's hospital with a diagnosis of sepsis, per International Classification of Diseases, Ninth Revision (ICD-9)¹⁶ and ICD Tenth Revision (ICD-10)17 codes (Supplemental Table S1), between July 1, 2013, and June 30, 2016. Patients were screened for matching ICD codes using the Duke Enterprise Data Unified Content Explorer (DEDUCE) patient data portal, which is a web-based query tool that allows access to requested data from the electronic medical record. For patients with a sepsis diagnosis, chart review, including body temperature, heart rate, respiratory rate, and white blood cell count, was then completed to determine the exact time when patients met SIRS criteria (Supplemental Table S2). All laboratory values used in the study were based on the first set of labs within 2 hours of sepsis diagnosis. The primary outcome for this project was in-hospital mortality. Secondary outcomes were assessed in relation to time from sepsis diagnosis, time zero starting when 2 or more SIRS criteria were met, and included time to mortality, hospital LOS, PICU admission, PICU LOS, intubation, time to intubation, time on the ventilator, and presence of end organ dysfunction at sepsis diagnosis (Supplemental Table S3).

Study Patients. Patients were included if they were age <18 years, had a diagnosis of sepsis per ICD codes, and met 2 or more age-based SIRS criteria with a confirmed or suspected source of infection (Supplemental Table S2). Patients were excluded if they were transferred from an outside hospital or facility or if they had a diagnosis of cystic fibrosis. The study was conducted at a children's hospital with 190 pediatric inpatient beds, including 16 PICU beds, 16 pediatric cardiac ICU beds, and 67 neonatal ICU beds. The institution also has a dedicated 18-bed pediatric ED.

Other Definitions. Appropriate antimicrobial administration was defined as receiving an antimicrobial agent to which all isolated pathogens were susceptible. If cultures were negative, therapy was deemed appropriate if the choice of empiric antibiotics was consistent with local guideline recommendations based on the presumed source and patient risk stratification via a sepsis order set at the institution (Supplemental Table S4). These antimicrobials also had to penetrate in adequate concentrations into the presumed source of infection. Patients were considered to be immunosuppressed if they had a history of bone marrow or cord blood transplant, solid organ transplant, malignancy, current or recent chemotherapy within the last 30 days, chronic systemic steroid use more than 7 days, or chronic use of immunosuppressive agents for more than 7 days.

Statistical Analysis. Descriptive statistics, including median and IQR, were used to describe continuous variables. Counts and percentages were used to describe categoric variables. Comparisons between continuous variables were conducted using the Kruskal-Wallis test. Comparisons between categoric variables were conducted using the χ² test. Statistical significance was defined as a p value <0.05. All data were analyzed using SAS version 9.4 (SAS Institute Inc, Cary, NC).

Results

Demographics. A total of 250 pediatric patients were screened for inclusion, of whom 181 were excluded. Reasons for exclusion included no documented confirmed or suspected source of infection (n = 131), transfer from an outside hospital (n = 37), did not meet sepsis criteria (n = 12), and other (n = 1). After application of the exclusion criteria, 69 patients (23.4%) were included in the analysis: 28 (40.6%) who received appropriate antibiotics within 1 hour of meeting sepsis criteria and 41 (59.4%) who received appropriate antibiotics more than 1 hour after meeting sepsis criteria.

Baseline characteristics were similar between the groups (Table 1). Patients were predominantly white males, with a median age of 0.9 years. The SIRS criteria at the time of meeting sepsis diagnosis did not differ between groups. Most patients had confirmed bacteremia or urinary tract infection as their source of infection. Overall, 19 patients (28%) were immunosuppressed, 10 in the greater than 1 hour group and 9 in the within 1 hour group.

Table 1.

Baseline Characteristics

Variable	Total (N = 69)	≤1 hr (n = 28)	>1 hr (n = 41)	p value
Age, median (IQR), yr	0.9 (0.1, 7.0)	0.9 (0.1, 12.5)	0.9 (0.1, 4.0)	0.7785
Weight, median (IQR), kg	8.9 (3.4, 20.4)	9.7 (3.5, 30.0)	8.5 (2.9, 17.6)	0.3192
Sex, male, %	65.2	71.4	61.0	0.3707
Ethnicity, %
White	42.0	42.9	41.5	0.9596
Black/African American	34.8	35.7	34.1
Other	23.2	21.4	24.4
SIRS criteria, median (IQR)
Body temperature, °C	37.2 (36.8, 38.6)	37.3 (36.7, 39.4)	37.0 (36.8, 38.4)	0.4332
RR, breaths per min	43.0 (27.0, 59.5)	38.0 (26.0, 60.0)	43.0 (28.0, 59.0)	0.4479
HR, beats per min	174.0 (143.0, 185.0)	149.0 (128.5, 191.0)	181.0 (147.0, 184.0)	0.5653
WBC, ×103 cells/μL	12.0 (7.0, 19.0)	12.0 (5.4, 16.8)	12.1 (7.4, 19.4)	0.8373

HR, heart rate; RR, respiratory rate; SIRS, systemic inflammatory response syndrome, at time of sepsis diagnosis; WBC, white blood cell count

Primary and Secondary Outcomes. In-hospital mortality was 7.1% (n = 2) in the within 1 hour group versus 14.6% (n = 6) in the greater than 1 hour group (p = 0.3399). Median hospital LOS was significantly shorter in the within 1 hour group versus the greater than 1 hour group (15.4 and 39.2 days, respectively, p = 0.0022). There was no difference between the groups for PICU admission or development of end organ dysfunction (Table 2). Overall, 72.5% (n = 50) of patients were admitted to the PICU, 36% (n = 18) of whom received appropriate antimicrobials within 1 hour of meeting sepsis criteria, and 64% (n = 32) who received appropriate antimicrobials greater than 1 hour after meeting sepsis criteria. Median PICU LOS was significantly shorter in the within 1 hour group versus the greater than 1 hour group (3.1 days and 33.6 days, respectively, p = 0.0191). There were no differences between groups for median time to intubation or median time spent on the ventilator.

Table 2.

Results

Variable	Total (N = 69)	≤1 hr (n = 28)	>1 hr (n = 41)	p value
In-hospital mortality, %	11.6	7.1	14.6	0.3399
Secondary end points
Hospital LOS, median (IQR), days	24.3 (11.9, 73.4)	15.4 (8.7, 26.3)	39.2 (16.1, 111.2)	0.0022
PICU admission, %	72.5	64.3	78.0	0.2088
End organ dysfunction, %	40.6	50.0	34.1	0.1879
	Total (n = 50)	≤1 hr (n = 18)	>1 hr (n = 32)
Secondary end points: PICU patients, median (IQR), days
PICU LOS	10.5 (2.1, 51.1)	3.1 (1.6, 12.7)	33.6 (3.6, 69.6)	0.0191
Time to intubation	0.0 (0.0, 2.7)	0.0 (0.0, 0.5)	0.1 (0.0, 3.7)	0.0705
Time on ventilator	5.1 (2.1, 20.5)	12.7 (2.1, 23.8)	5.1 (2.1, 20.5)	0.8949

LOS, length of stay; PICU, pediatric intensive care unit

Discussion

The Surviving Sepsis guidelines¹ advocate for administration of antimicrobials within 1 hour of diagnosis of sepsis or septic shock in both adult and pediatric patients. This recommendation is largely based on data from the adult patient population.^3–7 This study attempted to further address the outcomes of pediatric patients who receive antimicrobials within 1 hour of meeting sepsis criteria, given conflicting results that have been previously published in this patient population.^8–15 This study demonstrated pediatric patients who receive appropriate antimicrobial therapy within 1 hour of meeting sepsis criteria compared with greater than 1 hour after meeting sepsis criteria have improved outcomes, including shorter PICU and hospital LOS. Although non-significant, mortality in the within 1 hour group was half the mortality in the greater than 1 hour group.

Kumar et a¹³ studied more than 2000 adult patients with septic shock in their 2006 retrospective chart review. In this study, only 78.9% of patients received effective antimicrobial therapy after the onset of recurrent or persistent hypotension.³ Delay in effective antibiotics in this study was correlated with a significant increase in in-hospital mortality (95% CI, 1.103–1.136).³ Each hour delay in antimicrobial administration following the onset of persistent or recurrent hypotension was associated with a 7.6% decrease in survival.³ Compared with patients receiving appropriate therapy within the first hour, patients who received therapy after 1 hour of sepsis diagnosis had a significantly greater risk of in-hospital mortality (OR, 1.67; 95% CI, 1.12–2.48).³ This landmark trial is further supported by many other adult trials that have demonstrated mortality differences as well as lower incidence of progression to septic shock related to timely antibiotic administration.^4–7

Although pediatric data are less robust, there are several studies in this patient population that have explored the timing of antimicrobial administration in sepsis. One of the first pediatric studies to address this issue by Weiss and colleagues⁸ evaluated the impact of antimicrobial administration timing on mortality. In their retrospective observational study, risk of mortality increased with each hour delay in antibiotic administration from the time of sepsis diagnosis; however, this was not statistically significant until 3 hours.⁸ Patients with more than a 3-hour delay in both initial and first appropriate antimicrobial agents had a significant increase in mortality (OR, 3.92; 95% CI, 1.27–12.06; and OR, 3.59; 95% CI, 1.09–11.76, respectively).⁸ These results persisted when patients were stratified based on severity of illness.⁸ A more recent prospective observational study by Sankar and colleagues¹⁵ evaluated the risk of mortality in pediatric patients receiving antibiotics within 1 hour of sepsis diagnosis. In this study, most patients (77%) met criteria for septic shock, and delay in antibiotic therapy significantly increased risk of mortality (OR, 1.83; 95% CI, 1.14–2.92).¹⁵

Evans et al⁹ evaluated the impact of sepsis bundle completion within 1 hour on in-hospital mortality. In this study of 1179 patients, completion of the entire bundle within 1 hour, which included antibiotic administration, blood culture obtainment, and a fluid bolus, was also associated with lower in-hospital mortality (OR, 0.59; 95% CI, 0.38–0.93).⁹ Although these pediatric studies have mirrored results in the adult patient population, antibiotic timing in pediatric sepsis remains controversial because other large retrospective analyses have not found an association between time to antibiotic administration and improved outcomes in pediatric patients with sepsis, including mortality.^13,14

In the present study, patients in the within 1 hour group had significantly shorter hospital and PICU LOS. These end points have been evaluated in few other pediatric studies. Lane and colleagues¹³ did not find an association between antibiotic timing in pediatric patients with septic shock and PICU admission rate or hospital LOS. Similarly, conflicting results in another recent study found increased PICU and hospital LOS in children who received antibiotics within 1 hour from sepsis diagnosis.¹⁴ This has been further challenged by more recent data suggesting antibiotic administration within 1 hour increases hospital-free days.¹⁵ The findings from this more recent study as well as the present study demonstrate the impact that timing of antimicrobial administration could have on total hospital cost related to the LOS and that this outcome should be further explored. The PICU LOS is both reflective of patient severity of illness as well as an important measure of hospital resource use. Shorter PICU LOS may be related to faster time to defervescence in patients who received antibiotics within 1 hour. Reducing both PICU and hospital LOS is also essential for reducing risk of associated morbidities.

Although there were no statistically significant differences between the groups in our study for the development of end organ dysfunction, the absolute increase in the incidence of end organ dysfunction in the within 1 hour group may be attributed to a sicker patient population, although SIRS criteria did not differ between groups at baseline. Balamuth et al¹⁰ evaluated the effects of a protocolized sepsis bundle in pediatric patients in the ED, finding completion of the sepsis protocol was associated with the development of less organ dysfunction. In the study by Weiss and colleagues⁸ mentioned previously, patients who received initial antibiotics more than 3 hours after sepsis diagnosis had fewer organ failure–free days (16 versus 20 days, p = 0.04); however, other studies have not observed a relationship between faster antibiotic administration and less development of organ dysfunction.¹⁴ In addition, there was no difference in time to intubation in our study population, indicating patients were either intubated prior to meeting sepsis criteria or immediately afterwards. Time to intubation has not been evaluated in other pediatric studies; however, Sankar and colleagues¹⁵ demonstrated an increase in ventilator-free days in pediatric patients receiving antibiotics within 1 hour of sepsis diagnosis. There was no difference between groups for time on the ventilator; however, it is also important to note that time on the ventilator was longer in the within 1 hour group, despite shorter PICU LOS. This was influenced by several patients who were outliers and should be considered when interpreting the results of the PICU subgroup analysis.

Our study has several limitations. We chose to analyze the time between meeting sepsis criteria to time of appropriate antimicrobial administration, using SIRS criteria as a surrogate for sepsis onset. We also chose to use SIRS criteria because the Surviving Sepsis Guidelines recommend starting antimicrobial therapy within 1 hour of recognition of sepsis or septic shock.¹ The use of SIRS criteria as a surrogate for time of sepsis onset differs from previous studies that have used other time points, including time of presentation to the ED or triage time.^8,10,11 With SIRS criteria being a common method in clinical practice to aid in the diagnosis of sepsis, we believe this would be the most accurate determinant of sepsis recognition, especially in patients already admitted to the hospital. This, however, is limited, given the severity of illness at sepsis diagnosis was not assessed. The patient population included in this study is extremely diverse, including oncology, solid organ transplant, and bone marrow transplant populations. We believe this is the reason for longer LOS in comparison with other studies. It is also important to note that this may have confounded the difference found between groups for LOS, although the groups did not differ at baseline for number of immunocompromised patients included.

The diversity and complexity of the patients included in this study may limit the generalizability of the study results. In addition, the study was performed at a single center and includes a small patient population, which may explain the reason a significant difference was not found for the primary end point. Differences in local pathogen susceptibility and empiric antibiotics should also be taken into account because this may impact patient outcomes. Underlying chronic disease states were not assessed, which could account for differences found between groups. Another limitation of this study includes the retrospective nature of the study design. Some of the patients in this review were initiated on antimicrobials prior to meeting sepsis criteria. This indicates SIRS criteria do not include all components pediatric providers assess when deciding to initiate antimicrobial therapy in our patient population, and that use of meeting SIRS criteria as time zero in defining sepsis is a potential limitation. In addition, ICD codes were used to identify patients for inclusion, which may be insensitive for capturing all pediatric patients who likely met sepsis criteria within the institution.

Conclusion

In summary, this study supports administration of appropriate antimicrobial agents within 1 hour of recognition of sepsis in the pediatric patient as recommended by the Surviving Sepsis Guidelines.¹ The results strengthen a well-known observation across many patient populations that delaying antimicrobial therapy in sepsis or septic shock is associated with adverse outcomes.^2–12,15 Healthcare teams should continue to be diligent in recognizing sepsis, and processes should be streamlined in pediatric medical centers to improve any gaps in administering antimicrobials in a timely manner in the sepsis patient population. Larger studies are needed to determine if a mortality benefit exists when antimicrobials are administered within 1 hour of recognizing sepsis or septic shock in pediatric patients. Although a randomized design would limit selection bias, the ethical implications of administering antibiotics greater than 1 hour after sepsis or septic shock diagnosis would not allow for this study design.

ABBREVIATIONS

DEDUCE

Duke Enterprise Data Unified Content Explorer

ED

emergency department

LOS

length of stay

PICU

pediatric intensive care unit

SIRS

systemic inflammatory response syndrome