a b s t r a c t

Background: Early antibiotics are fundamental to sepsis management. Second-dose antibiotic delays were associ- ated with increased mortality in a recent study. Study objectives include: 1) determine factors associated with delays in second-dose antibiotic administration; 2) evaluate if delays influence clinical outcomes.

Methods: ED-treated adults (>=18 years; n = 1075) with severe sepsis or septic shock receiving >=2 doses of intra-

venous antibiotics were assessed, retrospectively, for second-dose antibiotic delays (dose time > 25% of recom- mended interval). Predictors of delay and impact on outcomes were determined, controlling for MEDS score, 30 mL/kg fluids and antibiotics within three hours of sepsis onset, lactate, and renal failure, among others.

Results: In total, 335 (31.2%) patients had delayed second-dose antibiotics. A total of 1864 second-dose antibiotics

were included, with 354 (19.0%) delays identified by interval (delayed/total doses): 6-h (36/67) = 53.7%; 8-h (165/544) = 30.3%; 12-h (114/436) = 26.1%; 24-h (21/190) = 8.2%; 48-h (0/16) = 0%. In-hospital mortality

in the timely group was 15.5% (shock-17.6%) and 13.7% in the delayed group (shock-16.9%). Increased odds of delay were observed for ED boarding (OR 2.54, 95% 1.81-3.55), shorter dosing intervals (6/8-h- OR 2.99, 95% CI 1.95-4.57; 12-h- OR 2.46, 95% CI 1.72-3.51), receiving 30 mL/kg fluids by three hours (OR 1.42, 95% CI 1.06-1.90), and renal failure (OR 2.57, 95% CI 1.50-4.39). Delays were not associated with increased mortality (OR 0.87, 95% CI 0.58-1.29) or other outcomes.

Conclusions: Factors associated with delayed second-dose antibiotics include ED boarding, antibiotics requiring more frequent dosing, receiving 30 mL/kg fluid, and renal failure. Delays in second-dose administration were not associated with mortality or other outcomes.

(C) 2021

1. Introduction

Antimicrobial therapy has been a mainstay in sepsis care. Early anti- biotics have been shown to improve patient survival [1-7] with current guidelines recommending administration of broad-spectrum antibiotics within one hour of patient presentation [8]. While studies have regu- larly highlighted the associated risks with time delays in first antibiotic administration for sepsis, to our knowledge, only one has evaluated the

* Corresponding author at: 800 University Bay Drive, Suite 300-56, Mail Code 9123, Madison, WI 53705, United States.

E-mail address: [email protected] (H.I. Kuttab).

timing of second-dose antibiotics. A retrospective analysis of 828 pa- tients demonstrated that significant delays in the timing of the second dose of antibiotics occurred in 33% of cases and were associated with in- creased mortality and need for mechanical ventilation [9]. Additionally, factors associated with timing delays were identified and included anti- biotics with more frequent dosing (i.e., 6 to 8 h) as well as emergency department (ED) boarding at the time of the second dose [9]. The phar- macodynamic impact of timing for the subsequent dosing of antibiotics has been shown to be important for clinical efficacy and the prevention of antibiotic resistance [10-12]. Therefore, delays with second-dose an- tibiotics may contribute to deleterious outcomes, especially as many empiric antibiotic regimens include dosing at shorter time intervals.

https://doi.org/10.1016/j.ajem.2021.03.057

0735-6757/(C) 2021

The rise in acute, unscheduled care in the United States has made ED overcrowding commonplace, in part, secondary to inPatient boarding [13]. As a result, critically ill patients subjected to prolonged ED boarding and delays to intensive care unit (ICU) transfer may be at risk for decreased compliance with Sepsis Bundles [14], higher mortal- ity, and longer duration of mechanical ventilation and ICU length of stay (LOS) [13,15-18]. Given that goals of sepsis care increasingly strive for early recognition and Rapid treatment [8], it stands to reason subse- quent dosing of antibiotics and other medications may be necessary while the patient remains in the ED.

There remains a need to identify interventions targeted at reducing in- hospital mortality for patients with sepsis. The objectives of our study are to: 1) examine the frequency and factors associated with delays in the ad- ministration of the second dose of antibiotics in a cohort of patients with severe sepsis and septic shock; 2) determine if significant delays influ- enced a variety of clinical outcomes, including in-hospital mortality, need for intubation or vasopressors, and length of stays.

1. Materials and methods
   1. Study design

This was a retrospective analysis from January 1, 2014 to April 30, 2015 and February 1, 2016 to May 31, 2017. Data identifying positive sepsis cases were originally collected as part of a quality improvement initiative and were not available between these dates. The cohort in- cluded adult patients (>=18 years old) presenting to the ED with severe sepsis or septic shock at an urban, tertiary-care center. Most patients treated in this ED are African American (~73% of all patients). The study was approved by the Institutional Review Board.

• 1. Patient selection

We aimed to adhere to the Strengthening the Reporting of Observa- tional Studies in Epidemiology (STROBE) guidelines [19]. We evaluated the study population for the presence of severe sepsis or septic shock within eight hours of ED arrival. Cases were identified by International Classification of Disease-Revision (ICD) 9 or 10 codes at hospital dis- charge related to sepsis and are the same as those used for case selection by the Centers for Medicare and Medicaid Services (Supplemen- tary Section 1). Next, to capture instances where there was clinical con- cern for infection, we flagged orders for blood cultures, urine cultures, or antibiotics within 12 h of ED arrival. A physician reviewed each chart using a standardized process to determine whether infection-related organ dysfunction occurred within eight hours of ED arrival (defined in Supplementary Section 2). Each case was confirmed with treating providers and a multidisciplinary group to improve quality control. These methods are similar to our group’s prior model for patient selec- tion [20].

Patients were included in the analysis if antibiotics were initiated within 12 h of sepsis onset, dosing occurred either in the ED or within four hours of leaving the ED, and the second dose was given within 48 h of the dose-time recommendation. Patients receiving a different antibiotic but within the same antibiotic class (e.g., first dose - ceftriax- one; second dose - cefepime) were included based on the recom- mended dose-timing of the first antibiotic.

• 1. Exclusion criteria

Patients who expired prior to receiving their second dose of antibi- otics, with left ventricular assist device (unless initial lactic acid >2.0 mmol/L), made do-not-resuscitate or comfort care within 12 h of sepsis onset, received IV antibiotics 24 h prior to presentation, transferred to the operating room within three hours of sepsis onset, or received a sec- ond dose of antibiotic that was from a different class (e.g., first dose- ceftriaxone; second dose-levofloxacin, n = 4) were excluded (Fig. 1).

• 1. Data abstraction and accuracy

Computational algorithms identified cases, demographic data, and a portion of clinical factors which assisted in blinding the selection pro- cess. Notably, the physician initially identifying eligible cases was not blinded, although these cases were prospectively selected as part of a quality improvement initiative and not for the purposes of this study. Additional sepsis-relevant interventions and clinical factors were man- ually collected. We aimed to adhere to the recommendations by Kaji et al. for reducing bias in chart review [21]. Abstractors were trained using strict definitions and protocols and were closely monitored by the principal investigator (MAW) to ensure data accuracy. Timing of an- tibiotics was obtained from the medication administration record, as opposed to the time at which they were ordered, so as to better reflect the time at which the antibiotic was actually initiated. A randomly se- lected subset of 75 charts were reviewed, with raw agreement of 91.4% and inter-rater reliability, ? = 0.83. Abstractors were not blinded to study purpose.

• 1. Relevant definitions

Relevant definitions for this study (i.e. sepsis classifications) can be found in Supplementary Section 2. Time of sepsis onset was deter- mined when all the criteria for severe sepsis/septic shock (Sepsis-2 criteria) were met and/or when laboratory values resulted. For the purpose of this study, second doses of antibiotics were considered to be ‘delayed’ if they occurred >25% outside the recommended range (e.g. for dose-timing recommendation of 12 h, a subsequent dose

>15 h was considered delayed) and mirrors the criteria outlined in Leisman et al. [9]. For antibiotics where a dosing range is considered ac- ceptable (i.e. 6-8 h), the longer interval was used for determining whether an antibiotic was delayed. For vancomycin in the setting of renal insufficiency or acute kidney injury, timing was based on drawn vancomycin levels approximately 24 h from first dose. All antibiotic dosing was evaluated and audited by a clinical pharmacist (NNP),

Fig. 1. Patient selection diagram.

multivariable regression for clinic”>who worked in conjunction with the principal investigator to determine whether antibiotic timing was appropriate. If disagreement occurred, a third-party adjudicated antibiotic timing appropriateness (BLW). For patients who received more than one antibiotic, dosing was determined to be delayed if any one of the antibiotics were delayed. All antibiotics were tracked, with exception of metronidazole or any agent not targeting bacteria (i.e., antifungals or antivirals). All tracked antibiotics and the corresponding recommended dosing guidelines based on renal function are listed in the Supplementary Section 3.

• 1. Statistical analysis

Statistical analysis was performed using SPSS Statistics 26 (IBM, Armonk, New York). Demographics and other baseline characteristics were summarized using means and standard deviations, medians and interquartile ranges, or frequency counts and percentages, as appropri- ate. Multivariable logistic regression identified the predictors for antibi- otic delays and impact of antibiotic delays on relevant clinical outcomes. Regression models evaluating for predictors of delays in second dosing included: ED boarding at time of second dose, 6-8, 12-, and 24-48-h dosing intervals, receiving 30 mL/kg of fluids (30by3) and antibiotics within 3 h of sepsis onset, ICU admission, renal failure, age, Mortality in Emergency Department Sepsis (MEDS) score, lactic acid, and pres- ence of acute kidney injury. Regression models evaluating for impact on clinical outcomes included delayed second dosing of antibiotics, MEDS score, receiving 30 mL/kg of fluids and antibiotics within 3 h of sepsis onset, lactic acid, renal and heart failure. ICU and hospital LOS were analyzed using Cox regression. Analysis for predictors of delay oc- curred at the antibiotic level, while analysis for outcomes occurred at the patient level (i.e. patients were considered to have a delayed second dose if any received antibiotic second-dose was delayed). Variables were selected a priori and based off prior analyses [9,20]. Statistical sig- nificance was defined as p < 0.05.

1. Results
   1. Demographics, outcomes, and sepsis measures

Demographic information, outcomes, and sepsis measures are outlined in Table 1. Groups were dichotomized into timely dose of second-dose antibiotics (timely) vs. delayed second-dose (delayed, see above). Notable differences between groups included a lower ED LOS (7.9 vs. 9.1 h) and lower rate of second dose of antibiotics given in the ED (14.6 vs. 38.5%) in the timely group. Additionally, patients in the timely group were less likely to achieve 30 mL/kg (30by3) (47.8 vs. 55.8%) and to receive antibiotics within three hours of sepsis onset (85.8 vs. 89.9%). Frequency of septic shock, mortality outcomes, and source of sepsis were similar between groups.

• 1. Frequency of delayed second dose

Frequency and median times to second dose are highlighted in Table 2, grouped by dosing interval for the second dose of antibiotic. Of the 1864 total second-dose antibiotics, delays were noted in 201 out of 611 (32.9%) in the 6-8 h dosing interval, 114 out of 436 (26.1%) in the 12-h dosing interval group, and 21 out of 206 (10.2%) in the 24-48 h interval group. Vancomycin that required lab-based dosing had 18 of 611 (2.9%) doses delayed per pharmacy adjudication.

• 1. Predictors of delayed second dose

Predictors of delayed second dose are outlined in Fig. 2. ED boarding at the time of the second dose (OR = 2.54, 95% CI 1.81-3.55), dosing at 6/8-h (OR = 2.99, 95% CI 1.95-4.57) and 12-h intervals (OR = 2.46, 95%

CI 1.72-3.51), achievement of 30by3 (OR = 1.42, 95% CI 1.06-1.90), and

history of end-stage renal disease (OR = 2.57, 95% CI 1.50-4.39) were associated with increased odds of delayed second dose.

• 1. Multivariable regression for clinical outcomes and subgroup analyses

After adjustment for potential confounders, the delayed second dose group did not demonstrate impact on mortality (OR = 0.87, 95% CI 0.58-1.29), intubation (OR = 0.71, 95% CI 0.47-1.07), need for vaso-

pressors (OR = 0.96, 95% CI 0.68-1.36), ICU LOS (HR = 1.08, 95% CI

0.89-1.31), or increased hospital LOS (HR = 1.00, 95% CI 0.87-1.16) (Table 3). In a subgroup analysis evaluating the 6-8 h dosing group, de- lays did not appear to impact mortality (OR 0.72, 95% CI 0.41-1.25) or

Table 1 Demographics, outcomes, sepsis measures. Table 1 highlights demographic, outcome, and sepsis measures between groups: timely second dose of antibiotics (timely) vs. delayed second dose of antibiotics (delayed, given >25% outside the recommended range). Abbre- viations: ED = emergency department, LOS = length of stay, ICU = intensive care unit, SIRS = systemic inflammatory response syndrome, qSOFA = quick sequential organ fail- ure assessment, IV = intravenous. Additional definitions are highlighted in the Supple- mentary Section 2. Beta-lactam antibiotics include penicillins, cephalosporins,

carbepenems, and monobactams
Variable	Total	Timely	Delayed
Total Patients	1075	740	335
Demographics
Age (years), mean +- SD	62.9 +- 17.3	63.9 +- 17.1	60.6 +- 17.5
Male sex, n(%)	542 (50.4)	353 (47.7)	189 (56.4)
Body Mass Index (kg/m2),	25.1	25.8	24.2
median (IQR)	(21.4-30.9)	(21.9-31.2)	(20.5-29.3)
Mortality in ED Sepsis Score,	10.0	10.0	11.0
median (IQR)	(8.0-14.0)	(7.3-14.0)	(8.0-14.0)
Time to diagnosis (hours),	1.3	1.3	1.0
median (IQR)	(0.4-2.2)	(0.5-2.3)	(0.3-2.1)
History of End-Stage Renal Disease,	111 (10.3)	72 (9.7)	39 (11.6)
n(%)
history of heart failure, n(%)	258 (24.0)	179 (24.2)	79 (23.6)
Septic Shock, n(%)	684 (63.6)	471 (63.7)	213 (63.6)
Outcomes Emergency Department LOS (hours),	8.3	7.9	9.1
median (IQR)	(6.2-11.1)	(6.0-10.6)	(6.8-11.9)
ICU admission, n(%)	632 (58.8)	432 (58.4)	200 (59.7)
ICU LOS (days), median (IQR)	1.0	1.0	1.0
	(0.0-3.0)	(0.0-3.0)	(0.0-3.0)
Hospital LOS (days), median (IQR)	7.4	7.6	7.2
	(4.3-12.8)	(4.2-12.9)	(4.6-12.8)
Mortality, n(%)	161 (15.0)	115 (15.5)	46 (13.7)
Shock (% of shock group)	119 (17.4)	83 (17.6)	36 (16.9)
Sepsis Measures
SIRS+, n(%)	923 (85.8)	639 (86.4)	284 (84.8)
qSOFA+, n(%)	371 (34.5)	254 (34.3)	117 (34.9)
Antibiotics within 3 h, n(%)	936 (87.1)	635 (85.8)	301 (89.9)
Time to first antibiotic (hours),	0.7	0.7	0.6
median (IQR)	(-0.3-1.8)	(-0.2-1.9)	(-0.4-1.5)
Second dose of antibiotics in ED, n(%)	237 (22.0)	108 (14.6)	129 (38.5)
30 mL/kg IV fluids in 3 h, n(%)	541 (50.3)	354 (47.8)	187 (55.8)
Lactate, mmol/L, median (IQR)	2.7	2.7	2.8
	(1.9-4.1)	(2.0-4.1)	(1.9-3.9)
Intubation, n (%)	153 (14.2)	114 (15.4)	39 (11.6)
Central line placed, n(%)	336 (31.3)	229 (30.9)	107 (31.9)
Vasopressor, n (%)	220 (20.5)	150 (20.3)	70 (20.9)
positive blood cultures, n (%)	347 (32.3)	237 (32.0)	110 (32.8)
Sepsis Source, n(%)
Pneumonia	311 (28.9)	217 (29.3)	94 (26.5)
Genitourinary	266 (24.7)	192 (25.9)	74 (22.1)
Abdominal	162 (15.1)	101 (13.6)	61 (18.2)
Skin/soft tissue	152 (14.1)	103 (13.9)	49 (14.6)
Indwelling intravenous catheter	59 (5.5)	40 (5.4)	19 (5.7)
Other/unknown	125 (11.6)	87 (11.8)	38 (11.3)
Antimicrobials Given, n(% of doses
given)
Vancomycin	805 (74.9)	739 (91.8)	66 (8.2)
Beta-lactams	1019 (94.8)	685 (67.2)	334 (32.8)
Ciprofloxacin	12 (1.1)	5 (41.7)	7 (58.3)
Levofloxacin	7 (0.7)	7 (100.0)	0 (0)
Daptomycin	21 (2.0)	12 (57.1)	9 (42.9)

Frequencies and timing of antibiotic delays. Table 2 notes the frequency of major delays, based on recommended dosing interval of second dose of antibiotics, highlighting significant time delays (i.e. % within recommended timing 24 h = 50% means 2nd antibiotic re-dosed at 36 h, and - 50% means 2nd antibiotic re-dosed at 12 h. *A total of 611 second doses of vancomycin administered per pharmacy recommendations based on laboratory levels (e.g. Vancomycin levels) with 18 delays. Therefore, the second dose of vancomycin administered based on lab- oratory values were delayed 2.9% (18 out of 611 occurrences) of the time

Dose Timing All Patients (n = 1075)

All Antibiotic doses (n = 1864)

Patients Delayed (n = 335) Doses Delayed (n = 354)

	n	Time to second dose (hr),	Median % within recommended timing	n	Time to second dose (hr),	% within recommended timing
		median (IQR)			median (IQR)
Second dose	67	7.9 (6.4-10.1)	32.0 (7.0-69.0)	36	10.1 (8.7-12.5)	68.0 (34.0-108.2)
at 6 h
Second dose	544	8.5 (6.9-10.7)	6.1 (-13.8-34.2)	165	12.4 (10.9-15.8)	55.4 (36.0-98.1)
at 8 h
Second dose	436	12.4 (10.7-16.5)	3.3 (-10.8-37.7)	114	22.7 (17.8-24.2)	89.0 (48.2-101.4)
at 12 h
Second dose	190	23.1 (15.0-24.9)	-3.8 (-37.5-3.8)	21	42.1 (36.7-49.7)	75.4 (52.9-107.1)
at 24 h
Second dose	16	24.2 (20.1-32.3)	-49.6 (-58.1 - -32.7)	0	-	-
at 48 h

other clinical outcomes (Supplementary Table 6). In a subgroup analysis evaluating delays specifically in patients with positive blood cultures, delays did not appear to impact mortality (OR 0.74, 95% CI 0.33-1.63) or other clinical outcomes (Supplementary Table 7).

1. Discussion

The impact of delays in antibiotics in septic patients has been the subject of considerable interest, with various sources demonstrating substantial Mortality benefits to early initiation of antimicrobials [1-6]. To our knowledge, only one study has evaluated for predictors of second-dose antibiotic delays and the impact this may have on clinical outcomes [9]. Our study demonstrates delays in the second dose of an- tibiotics were common. Furthermore, predictors of delays can be identi- fied, with ED boarding, antibiotics requiring more frequent dosing, and history of end-stage renal disease being associated with delays. How- ever, major delays were not found to be associated with mortality, intu- bation, need for vasopressors, or length of stays.

boarding of patients in the ED is, perhaps unsurprisingly, associated with delays in indicated care and has been associated with multiple ad- verse outcomes, including decreased compliance with sepsis bundles, mortality and hospital LOS [13-18]. The challenges that ED boarding presents for coordination of care are myriad, ranging from delays in the receipt of appropriate care to failures in inter-team/provider

communication. To this end, solutions that address the challenge of prolonged stays in the ED, especially after a patient has been accepted to an inpatient team, will likely reduce the risk of delays in care.

Shorter dosing intervals have been previously associated with in- creased odds of second-dose antibiotic delays in the context of community-acquired pneumonia in patients admitted from the ED [22]. The authors of this study suggest that antibiotic delays might be mitigated by emergency physicians selecting antibiotics with longer redosing intervals. For example, if attempting to cover for gram- positive organisms, selection of ceftriaxone, with a dosing frequency of every 24 h, rather than cefazolin which must be dosed every 6-8h in patients with normal renal function, may be beneficial. The specific pharmacodynamic impacts of antibiotic timing and achievement and maintenance of appropriate concentrations has also been shown to be important for clinical efficacy as well as the prevention of antibiotic re- sistance [10-12]. Thus, selection of antibiotics with longer dosing inter- vals would, in theory, limit periods of sub-optimal antibiotic concentration. There may be potential benefits for the prescription of antibiotics with longer dosing intervals as long as the necessary spec- trums of antibiotic activity were not compromised.

The prior analysis by Leisman et al. highlights the relatively common

frequency in major delays in second dose of antibiotics, with major de- lays occurring more frequently with antibiotics requiring more frequent dosing [9]. It was also demonstrated that major delays were associated

Fig. 2. Predictors of delayed second dose.

Table 3

multivariable regression analysis for clinical outcomes. Table 3 represents impact of a variety of measures on clinical outcomes, including delayed second dose of antibiotics. Ultimately, delayed second dose of antibiotics did not demonstrate reduction in mortality, intubation, vasopressor use, or length of stays. Abbreviations: 30 mL/kg in 3 h = 30 mL/kg of intravenous fluids by bolus given in the first three hours of sepsis presentation, ED = emergency department

Variable Mortality Intubation Vasopressors ICU LOS Hospital LOS

			OR	95% CI		OR	95% CI		OR	95% CI		HR	95% CI		HR	95% CI
	Delayed Group		0.87	0.58-1.29		0.71	0.47-1.07		0.96	0.68-1.36		1.08	0.89-1.31		1.00	0.87-1.16
	30 mL/kg in 3 h		0.65	0.44-0.95		0.86	0.59-1.27		1.27	0.90-1.79		1.48	1.21-1.81		1.22	1.05-1.40
	End-Stage Renal Disease		1.09	0.60-2.00		1.09	0.60-1.98		1.60	0.97-2.63		0.99	0.74-1.32		0.82	0.66-1.03
	Heart Failure		1.21	0.80-1.82		1.20	0.79-1.84		1.96	1.36-2.81		0.97	0.78-1.21		0.81	0.69-0.96
	Mortality in ED Sepsis Score		1.16	1.11-1.20		1.06	1.01-1.10		1.12	1.08-1.16		0.96	0.94-0.98		0.97	0.96-0.99
	Lactate		1.18	1.11-1.25		1.26	1.18-1.34		1.17	1.11-1.24		0.97	0.94-1.00		0.96	0.92-0.99
	Antibiotics by 3 h		1.25	0.74-2.10		1.18	0.69-2.02		1.33	0.83-2.13		0.95	0.72-1.25		1.02	0.83-1.24

with increased odds of mortality and mechanical ventilation; however, the group cautions interpretation, and suggests that perhaps delays in antibiotics may be a surrogate for patients who generally received less attention and care overall, given that many of the delays occurred in the ED [9]. Our study confirms several of these findings, including sim- ilar rates of antibiotic delays (31% in this cohort vs. 33% in the Leisman cohort) and ED boarding at the time of second dose (22% in both stud- ies) resulting in increased odds for second-dose delays (our cohort: OR 2.54, 95% CI 1.81-3.55 vs. Leisman: OR 2.67, 95% 1.74-4.09).

In contrast, this study does not demonstrate an association between second-dose antibiotic delays and clinical outcomes. Several key dis- tinctions between our study and the earlier study may account for these differences. First, our population demonstrated a higher propor- tion of patients admitted to the ICU compared to the Leisman group (58.8% vs. 26.3%, respectively), and perhaps patients in our cohort had more closely monitored care. Next, our cohort had earlier time to first antibiotic (~42 vs 92 min) and greater volumes of intravenous fluids ad- ministered (~2.1 vs 1.5 L) compared to the Leisman cohort, respectively. Perhaps this more aggressive, early care altered the differences in pa- tient outcomes. Additionally, it is possible that institutional differences may account for our discrepant clinical outcomes. This may include: the intermittent presence of a pharmacist in our ED who may help verify dosing of antibiotics and future dosing of vancomycin based on drug levels, variation in ED LOS, nursing-to-patient ratios (generally, 1-to- 4 at our institution, though this is highly variable based on the day), and timing of the assumption of patient care or releasing of inpatient or- ders following handoff to inpatient teams, even if the patient boards in the ED while awaiting an inpatient bed.

Within our specific cohort, there was also a greater achievement of 30by3 in the delayed group (55.8% vs. 47.8%), which may also have had significant impacts on clinical outcomes. Our prior analysis in a sim- ilar cohort of patients demonstrated a significant reduction in mortality in patients who reached 30 mL/kg of fluids within three hours of sepsis onset, without any significant mortality benefit noted with time to first- dose antibiotics [20]. This may be because the overall time to antibiotics in this population was relatively short (~42 min of sepsis identification), and perhaps impacts in mortality are seen with more significant delays in antibiotics. A recent meta-analysis suggests that the 3-h time-mark may be the threshold in which mortality may significantly differ, and 87.1% of patients in our cohort received antibiotics in <3 h [23]. It may be that early, aggressive sepsis care alters the impact of more downstream care. We did attempt to control for these differences by in- cluding both initial antibiotic timing and compliance with fluid resusci- tation goals in our logistic regression models.

Lastly, both studies observed significant delays in the second dose of antibiotics in those compliant with initial sepsis care. Leisman et al. sug- gests that the time interval between first dose and the hand-off between ED and inpatient teams is inherently longer when initial antibiotics are administered earlier in the ED course [9]. Our study may validate this, with an Average ED LOS of ~8 h, suggesting that handoffs and transport to Inpatient units may occur at the time at which the second dose of

antibiotics may be due, especially as the majority of delays observed in both studies occurred at the 6-8 h dosing interval. Leisman et al. also suggests that successful Initial resuscitation efforts could lead to im- provement in outcomes and admission to the Inpatient floor, whereas noncompliant resuscitation may result in patients being sent to the ICU [9]. Our study may also validate this, with increased rates of delayed hypotension (66.6% vs. 63.9%) in the delayed antibiotic group. It is also feasible that physician practices regarding sepsis care are subject to “cognitive off-loading”. That is, following the successful resuscitation of a septic patient with intravenous fluids, antibiotics, vasopressors, etc., the physician feels that their management of the patient is other- wise “complete,” and may defer the ordering of subsequent antibiotics to inpatient providers. This may be the result of educational initiatives often placing emphasis on early, aggressive resuscitation of patients with sepsis, with little attention being given to the downstream effects of delays in sepsis care. However, the practice patterns of physicians in both cohorts are extremely difficult to ascertain, especially in a retro- spective fashion and across differing institutions. Perhaps interventions designed at improving communication between ED and inpatient pro- viders or improvements in sepsis bundles, including automated order- ing of subsequent antibiotic doses, may be worthwhile.

This study has several key limitations. First, this study cannot deter- mine any causal relationships as it is the experience of a single medical center and conducted retrospectively. Various sepsis initiatives were ongoing during the study time period and, therefore, it is difficult to as- certain whether these initiatives had any impact on study outcomes. Another potential limitation of the present study is the discontinuity of the study period. This data was obtained from a sample across two time periods separated by just under a year, but the data missing from this time cannot be obtained. Additionally, changes in practice pattern across the study period, i.e. related to the implementation of CMS guide- lines related to the SEP-1 measure, present additional challenges to Data interpretation [24]. As these changes were being implemented, it is con- ceivable that changes in baseline mortality might also have occurred, providing a potential confounder. An additional confounder exists given that teams may have held or discontinued antibiotics during ad- mission, then later re-ordered for another reason; at our institution, in- patient teams assume care of patients boarding in the ED, and it is up to ED nurses to implement the orders. However, the timing of when orders were placed or released are likely highly variable. We attempted to con- trol for this by including patients who received their second dose of an- tibiotics within 48 h of the recommended intervals. Further, there is an inherent difficulty in diagnosing sepsis; included patients had vital sign derangements, laboratory findings, and signs and symptoms consistent with sepsis. Included patients met diagnostic criteria as it existed at the time, and as indicated in Table 1, most patients had an identified source of infection at time of discharge, suggesting that our cohort, in fact, rep- resents patients with severe sepsis and septic shock, rather than other conditions triggering a systemic inflammatory response without infec- tious trigger. Moreover, given the relatively small sample size and the even smaller rates of exposure (second-dose delay) and outcome (in-

hospital death), our study might be underpowered to identify an out- come difference, should it exist. It should be noted that the present study is larger than the comparable study by Leisman et al. Lastly, inher- ent differences exist between our study groups. The delayed group had a reduced time to sepsis diagnosis and achieved greater frequency of reaching 30 mL/kg of fluids; however, we attempted to control for these differences, among others.

1. Conclusions

Delays in second-dose antibiotics in patients with sepsis are com- mon and associated with ED boarding and antibiotic dosing frequency. However, no impact of mortality or other clinical outcomes was deter- mined in this cohort of patients. Further studies are required on this topic, especially since previous work had observed deleterious effects and rate of delays are high.

Financial support

This project was supported by the National Center for Advancing Translational Sciences (NCATS) of the National Institutes of Health (NIH) through Grant Number 5UL1TR002389-02 that funds the Insti- tute for Translational Medicine (ITM). The content is solely the respon- sibility of the authors and does not necessarily represent the official views of the NIH.

Credit author statement Joseph Lykins, Hani Kuttab, Michael Ward: Conceptualization, For-

mal Analysis, Data Curation, Writing, Investigation, Writing - Original

draft and review & editing, Visualization, Supervision.

Erron Rourke, Michelle Hughes, Eric Keast, Jason Kopec: Data Curation, Validation.

Natasha Pettit, Brooke Ward: Investigation, Data Curation.

Declaration of Competing Interest

The authors have no conflicts of interest or competing interests to disclose.

Acknowledgements

We would like to thank David P. Liedke for his continued integral contribution to our work.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi. org/10.1016/j.ajem.2021.03.057.

References

1. Seymour CW, Gesten F, Prescott HC, Friedrich ME, Iwashyna TJ, Phillips GS, et al. Time to treatment and mortality during mandated emergency care for sepsis. New England Journal of Medicine. 2017;376:2235-44. https://doi.org/10.1056/ NEJMoa1703058.
2. Puskarich MA, Trzeciak S, Shapiro NI, Arnold RC, Horton JM, Studnek JR, et al. Asso- ciation between timing of antibiotic administration and mortality from septic shock in patients treated with a quantitative Resuscitation protocol. Crit Care Med. 2011; 39:2066-71. https://doi.org/10.1097/CCM.0b013e31821e87ab.
3. Kumar A, Roberts D, Wood KE, Light B, Parrillo JE, Sharma S, et al. Duration of hypo- tension before initiation of effective antimicrobial therapy is the critical determinant of survival in human septic shock*. Crit Care Med. 2006;34:1589. https://doi.org/10. 1097/01.CCM.0000217961.75225.E9.
4. Zasowski EJ, Bassetti M, Blasi F, Goossens H, Rello J, Sotgiu G, et al. A systematic re- view of the effect of delayed appropriate antibiotic treatment on the outcomes of pa- tients with severe bacterial infections. Chest. 2020. https://doi.org/10.1016/j.chest. 2020.03.087.
5. Liu VX, Fielding-Singh V, Greene JD, Baker JM, Iwashyna TJ, Bhattacharya J, et al. The timing of early antibiotics and hospital mortality in sepsis. Am J Respir Crit Care Med. 2017;196:856-63. https://doi.org/10.1164/rccm.201609-1848OC.
6. Seymour CW, Kahn JM, Martin-Gill C, Callaway CW, Yealy DM, Scales D, et al. Delays from first medical contact to antibiotic administration for sepsis. Crit Care Med. 2017;45:759-65. https://doi.org/10.1097/CCM.0000000000002264.
7. Sherwin R, Winters ME, Vilke GM, Wardi G. Does early and appropriate antibiotic administration improve mortality in emergency department patients with severe sepsis or septic shock? J Emerg Med. 2017;53:588-95. https://doi.org/10.1016/j. jemermed.2016.12.009.
8. Levy MM, Evans LE, Rhodes A. The surviving sepsis campaign bundle: 2018 update. Intensive Care Med. 2018;44:925-8. https://doi.org/10.1007/s00134-018-5085-0.
9. Leisman D, Huang V, Zhou Q, Gribben J, Bianculli A, Bernshteyn M, et al. Delayed sec- ond dose antibiotics for patients admitted from the emergency department with Sepsis: prevalence, risk factors, and outcomes. Crit Care Med. 2017;45:956-65. https://doi.org/10.1097/CCM.0000000000002377.
10. Roberts JA, Paul SK, Akova M, Bassetti M, De Waele JJ, Dimopoulos G, et al. DALI: de- fining antibiotic levels in intensive care unit patients: are current ?-lactam antibiotic doses sufficient for critically ill patients? Clin Infect Dis. 2014;58:1072-83. https:// doi.org/10.1093/cid/ciu027.
11. McKinnon PS, Paladino JA, Schentag JJ. Evaluation of area under the inhibitory curve (AUIC) and time above the minimum inhibitory concentration (T>MIC) as predic- tors of outcome for cefepime and ceftazidime in serious bacterial infections. Int J Antimicrob Agents. 2008;31:345-51. https://doi.org/10.1016/j.ijantimicag.2007.12. 009.
12. Roberts JA, Kruger P, Paterson DL, Lipman J. Antibiotic resistance-what’s dosing got to do with it? Crit Care Med. 2008;36:2433-40. https://doi.org/10.1097/CCM. 0b013e318180fe62.
13. Boarding Task Force ACEP. Emergency department crowding: High-impact solu- tions; 2016.
14. Shin TG, Jo IJ, Choi DJ, Kang MJ, Jeon K, Suh GY, et al. The adverse effect of emergency department crowding on compliance with the resuscitation bundle in the manage- ment of severe sepsis and septic shock. Crit Care. 2013;17:R224. https://doi.org/10. 1186/cc13047.
15. Mohr NM, Wessman BT, Bassin B, Elie-Turenne M-C, Ellender T, Emlet LL, et al. Boarding of critically ill patients in the emergency department. Journal of the American College of Emergency Physicians Open. 2020;1:423-31. https://doi.org/ 10.1002/emp2.12107.
16. Singer AJ, Jr HCT, Viccellio P, Pines JM. The association between length of emergency department boarding and mortality. Acad Emerg Med. 2011;18:1324-9. https://doi. org/10.1111/j.1553-2712.2011.01236.x.
17. Al-Qahtani S, Alsultan A, Haddad S, Alsaawi A, Alshehri M, Alsolamy S, et al. The as- sociation of duration of boarding in the emergency room and the outcome of pa- tients admitted to the intensive care unit. BMC Emerg Med. 2017;17:34. https:// doi.org/10.1186/s12873-017-0143-4.
18. Boulain T, Malet A, Maitre O. Association between long boarding time in the emer- gency department and hospital mortality: a single-center propensity score-based analysis. Intern Emerg Med. 2020;15:479-89. https://doi.org/10.1007/s11739-019- 02231-z.
19. von Elm E, Altman DG, Egger M, Pocock SJ, Gotzsche PC, Vandenbroucke JP. The strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. Ann Intern Med. 2007; 147:573-7. https://doi.org/10.7326/0003-4819-147-8-200710160-00010.
20. Kuttab HI, Lykins JD, Hughes MD, Wroblewski K, Keast EP, Kukoyi O, et al. Evaluation and predictors of fluid resuscitation in patients with severe Sepsis and septic shock. Crit Care Med. 2019;47:1582-90. https://doi.org/10.1097/CCM.0000000000003960.
21. Kaji AH, Schriger D, Green S. Looking through the retrospectoscope: reducing bias in emergency medicine chart review studies. Ann Emerg Med. 2014;64:292-8. https:// doi.org/10.1016/j.annemergmed.2014.03.025.
22. Shah MN, Schmit J, Croley WC, Meltzer D. Continuity of antibiotic therapy in patients admitted from the emergency department. Ann Emerg Med. 2003;42:117-23. https://doi.org/10.1067/mem.2003.257.
23. Sterling SA, Miller WR, Pryor J, Puskarich MA, Jones AE. The impact of timing of an- tibiotics on outcomes in severe sepsis and septic shock: a systematic review and meta-analysis. Crit Care Med. 2015;43:1907-15. https://doi.org/10.1097/CCM. 0000000000001142.
24. Centers for Medicare & Medicaid Services: Joint Commission. Specifications manual for national hospital inpatient quality measures; 2018.