Emergency Medicine

Association between emergency department sepsis order set design and delay to second dose piperacillin-tazobactam administration

a b s t r a c t

Background: Delay to first antibiotic dose in patients with sepsis has been associated with increased mortality. Second dose antibiotic delay has also been linked to worsened patient outcomes. Optimal methods to decrease second dose delay are currently unclear. The primary objective of this study was to evaluate the association be- tween updating an emergency department (ED) sepsis order set design from one-time doses to scheduled anti- biotic frequencies and delay to administration of second piperacillin-tazobactam dose.

Methods: This retrospective cohort study was conducted at eleven hospitals in a large, integrated health system and included adult patients treated in the ED with at least one dose of piperacillin-tazobactam ordered through an ED sepsis order set over a two year period. Patients were excluded if they received less than two doses of piperacillin-tazobactam. Midway through the study period, the enterprise-wide ED sepsis order set was updated to include scheduled antibiotic frequencies. Two patient cohorts receiving piperacillin-tazobactam were com- pared: those in the year before the order set update and those in the year post-update. The primary outcome was major delay, defined as an administration delay >25% of the recommended dosing interval, which was eval- uated with multivariable logistic regression and interrupted Time series analysis.

Results: 3219 patients were included: 1222 in the pre-update group and 1997 in the post-update group. The pro- portion of patients who experienced major second dose delay was significantly lower in the post-update group (32.7% vs 25.6%, p < 0.01; adjusted OR 0.64, 95% CI 0.52 to 0.78). No between-group difference was detected in the slope of monthly major delay frequency, but there was a significant level change (post-update change -10%, 95% CI -17.9% to -1.9%).

Conclusions: Including scheduled antibiotic frequencies in ED sepsis order sets is a pragmatic mechanism to decrease delays in second antibiotic doses.

(C) 2023

  1. Introduction

* Corresponding author at: Department of Pharmacy, Cleveland Clinic Main Campus, Hb-105, 9500 Euclid Ave, Cleveland, OH 44195, USA.

E-mail address: [email protected] (R.M. Erickson).

1 Present Address: Quality and Patient Safety, NewYork Presbyterian Hospital, New York, NY, USA.

2 Present Address: Department of Emergency Medicine, Columbia University Vagelos College of Physicians and Surgeons, New York, NY, USA.

Delay to initiation of antibiotics in patients with sepsis has been as- sociated with increased mortality. In patients with septic shock, the odds of mortality increase by 7% with every additional hour of delay to first dose antibiotic administration [1]. Although less frequently stud- ied, recent literature has reported 13 to 33% of patients experienced major delay to administration of second dose antibiotics [2-9]. Major delay to second dose antibiotic administration has been associated with increased mortality, length of stay, and duration of mechanical ventilation [2,4,7-9]. While factors associated with delay to second dose antibiotic administration have been identified, there is a paucity

https://doi.org/10.1016/j.ajem.2023.01.057

0735-6757/(C) 2023

of evidence supporting actionable interventions to decrease this delay in patients with sepsis [10]. We sought to evaluate the association be- tween updating an emergency department (ED) sepsis order set to in- clude scheduled antibiotic frequencies and delay to second dose piperacillin-tazobactam administration and clinical outcomes.

  1. Study design and methods
    1. Study design and population

This study was a retrospective pre-post cohort study of adult ED pa- tients with suspected or confirmed sepsis who received at least one dose of piperacillin-tazobactam ordered through an enterprise ED sep- sis order set between May 7, 2018 and May 6, 2020. Patients from eleven hospitals in our large, integrated health system, including 10 community hospitals and one large academic medical center, were eli- gible for study inclusion. Patients were excluded if they received less than two doses of piperacillin-tazobactam, received a second piperacillin-tazobactam dose >24 h after the first dose, or were trans- ferred to a different hospital from the ED (due to potential confounding timing delays and because orders do not continue between locations). The Cleveland Clinic Institutional Review Board reviewed and approved this study; a waiver of informed consent was granted.

The pre-update period was defined as the 12-month period from May 7, 2018 through May 6, 2019 when all antimicrobial doses ordered from the ED sepsis order set were defaulted to an ordered frequency of “once”, which is standard in ED order sets. In an attempt to mitigate delays to second dose antibiotic administration, the default order frequencies in this sepsis order set were updated to scheduled antibiotic frequencies with piperacillin-tazobactam orders defaulting to a fre- quency of every six hours beginning on May 7, 2019. The post-update period was defined as the subsequent 12-month period from May 7, 2019 through May 6, 2020.

In the majority of our patients, renal function labs have not resulted at the time of antibiotic ordering or verification. Due to expanded vol- ume of distribution in patients with sepsis and patients undergoing fluid resuscitation, aggressive initial beta-lactam dosing may be war- ranted in this population. Additionally, in one large database study, 57% of patients with infection had “transient acute kidney injury”, with resolution in 48 h [11]. Because of the wide safety margin of beta-lactams, in severely ill patients with renal dysfunction, our clini- cians commonly continue piperacillin-tazobactam at a dose of 3.375 g every 6 h for 24 to 48 h or until the trajectory of patients’ renal dysfunc- tion is clear. For less critically ill patients, decentralizED pharmacists re- view all renally adjusted antibiotics every morning for renal dose adjustments if indicated. For patients in both pre- and post-update groups, clinicians were able to modify default frequency for antibiotic orders based on clinical discretion.

    1. Outcomes

The primary objective of this study was to compare the proportion of ED patients who had a major delay to second dose piperacillin- tazobactam administration between pre- and post-update groups. The secondary objectives of the study were to evaluate the association between ED sepsis order set update and time to second dose piperacillin-tazobactam as well as in-hospital mortality and morbidity. Major delay was defined as a delay >25% of the recommended dosing interval between the first and second dose to align with previ- ously published literature [2-9]. The recommended dosing interval was determined using initial assessment of renal function in the ED and health-system guidelines based on creatinine clearance for piperacillin-tazobactam as follows: 3.375 g every 6 h for creatinine clearance >40 mL/min, 3.375 g every 8 h for creatinine clearance 20

to 40 mL/min, and 3.375 g every 12 h for creatinine clearance <20

mL/min or hemodialysis [12]. All piperacillin-tazobactam doses are in- fused as intermittent infusions over 30 min within our health-system.

In an a priori-planned subgroup analysis, study outcomes were also assessed in patients who met the Centers for Disease Control (CDC) Adult Sepsis Event (ASE) criteria [13,14]. This subgroup analysis was conducted to examine the association of order set update specifically in the intended target patient population adjudicated as having sepsis. The ASE definition was used in this analysis as it has the highest speci- ficity for infection of any chart-abstracted measure [15]. CDC criteria to meet the ASE definition include the following: blood cultures ob- tained, administration of at least four days of broad-spectrum antibi- otics (or until time of death, if sooner) initiated within two days of blood cultures, and evidence of organ dysfunction [13,14]. An additional subgroup analysis of patients with septic shock, defined as patients on vasopressors, was also conducted to examine the impact of order set up- date in a critically ill patient population. Furthermore, a sensitivity anal- ysis using an alternative definition for major second dose delay was also performed in which the actual ordered frequency for the second dose was utilized as the reference to assess major delay instead of using rec- ommended dosing frequency based on assessment of the patient’s renal function.

    1. Statistical analysis

Data were queried and abstracted from the electronic medical re- cord (EPIC Systems, Verona, WI) by a trained informatics specialist and included demographic and clinical information: severity of illness scores, piperacillin-tazobactam order data, laboratory values, patient disposition, mechanical ventilation details, and vasopressor use. Cate- gorical variables were analyzed using the Pearson chi-square and Fish- er’s exact tests and continuous variables using the Wilcoxon rank-sum and Student’s t-tests. A p-value of <0.05 was considered statistically sig- nificant.

Multivariable logistic regression models were developed to evaluate

the association between order set update and major delay, as well as the association between order set update and in-hospital mortality. Covari- ates included in the logistic regression models were determined a priori and selected based on prior literature and plausibility for association with the outcomes of interest [16]. The final variables included in the major delay logistic regression model were baseline Sequential Organ Failure Assessment score, ASE Sepsis criteria, initial ED lactate, second piperacillin-tazobactam dose due in the ED, second piperacillin-tazobactam dose due within 1 h of transfer from the ED, age, gender, ED disposition, and hospital. Variables included in the mor- tality logistic regression model included baseline SOFA score, ASE sepsis criteria, Charlson Comorbidity Index, vasopressor use, mechanical ven- tilation, piperacillin-tazobactam administration within 1 h of lactate collection, and ED disposition. A multivariable logistic regression model using the same covariates was also built for the sensitivity anal- ysis using an alternative definition for major second dose delay. To eval- uate the impact of the ED sepsis order set update on the proportion of patients with delay to administration of second dose piperacillin- tazobactam over time, quasi-experimental analyses utilizing interrupted time series analyses (also known as segmented regression) were conducted [17]. This method uses aggregate data collected over equally spaced time intervals to compare differences in slopes between regression segments both before and after order set design update and evaluates the change in level of the outcome (proportion of patients who experienced a delay in second dose piperacillin-tazobactam ad- ministration) immediately after the order set design was updated com- pared with the end of the period before the order set was updated (regression segment) as well as the change in slope before and after the order set was updated. This analysis is recommended for evaluation of longitudinal data sets when comparing pre/post-update outcomes and accounts for pre-update slope trends [18-20]. To conduct the interrupted time series analysis model, an ordinary least-squares

model with Newey-West standard error was utilized. Autocorrelation was assessed with the Cumby-Huizinga general test and non- stationarity was assessed with the modified Dickey-Fuller t-test. All analyses were conducted using STATA(R) (version 13.1, College Station, Texas) [17].

  1. Results
    1. Patient characteristics

A total of 4936 patients were screened for enrollment in the study and 1717 were excluded, with the most common reason for exclusion being no second dose of piperacillin-tazobactam administered (n = 886). Of the 3219 patients included, there were 1222 patients in the pre-update group and 1997 patients in the post-update group (Fig. 1). Patient characteristics are summarized in Table 1. Patients in the pre- update group had higher SOFA scores (4.0 vs 3.4, p < 0.01) and higher proportions of patients meeting ASE criteria (34.9% vs 28.4%, p < 0.01) and receiving first antibiotic doses within one hour of lactate collection (72.8% vs 64.3%, p < 0.01). There were no differences detected between groups in other patient characteristics.

    1. Outcomes

The proportion of patients who experienced a major delay in second dose of piperacillin-tazobactam administration was significantly lower in the post-update group (32.7% vs 25.7%, absolute difference -7% [95% CI -3.7% to -10.3%]) (Table 2). Further, time to second dose piperacillin-tazobactam administration was shorter in the post-update group (7.6 [3.8] vs 7.1 [2.9] hours). Duration of mechanical ventilation

was also shorter in the post-update group (5.6 [7.0] vs 4.2 [4.6] days, p = 0.02). There were no differences detected in hospital length of stay or in-hospital mortality between groups.

In the multivariable logistic regression model, the ED sepsis order set update was associated with a significantly lower odds of major delay (OR 0.66, [95% CI 0.55 to 0.81]) (Table 2). The order set update was not associated with in-hospital mortality (OR 1.02, [95% CI 0.71 to 1.47]). After the ED sepsis order set was updated, there was a significant level change in the monthly proportion of patients who experienced a major delay in their second dose of antibiotics: -10% [95% CI -18% to

-2%], however, there was no difference in the slope of the monthly pro- portion of patients who experienced a major delay in their second dose of antibiotics (change in slope: 0.03% [95% CI -0.87 to 0.93]) (Fig. 2).

Table 1

Patient characteristics.

Demographic Information

Pre-Update

Post-Update

p-value

Group

Group

(n = 1222)

(n = 1997)

Age, years

66.3 (16.1)

65.3 (17.0)

0.09

Male, n (%)

672 (55.0)

1098 (55.0)

>0.99

Weight, kg

83.8 (26.3)

84.6 (27.0)

0.44

Race, n (%)

0.54

White

759 (62.1)

1253 (62.7)

African American

397 (32.5)

620 (31.1)

Othera

45 (3.7)

79 (3.9)

Unavailable

21 (1.7)

45 (2.3)

Initial ED lactate

2.7 (2.0)

2.8 (2.2)

0.32

Maximum lactate in 24 h

3.0 (2.2)

3.0 (2.4)

0.66

Maximum lactate in 48 h

3.0 (2.3)

3.0 (2.4)

0.97

Antibiotics within 1 h of lactate, n (%)

698 (72.8)

760 (64.3)

<0.01

Blood culture collected, n (%)

1179 (96.5)

1921 (96.2)

0.68

First creatinine, mg/dL

1.83 (2.07)

1.79 (2.04)

0.60

Creatinine clearance, mL/min

64.5 (51.1)

66.2 (48.0)

0.35

Recommended piperacillin-tazobactam

0.32

dosing interval, hours; n (%)

Every 6 h

775 (63.4)

1319 (66.0)

Every 8 h

281 (23.0)

425 (21.3)

Every 12 h

166 (13.6)

253 (12.7)

Charlson Comorbidity Index

2.5 (2.5)

2.6 (2.7)

0.31

SOFA score

4.0 (3.4)

3.4 (3.2)

<0.01

Patients meeting ASE criteria, n (%)

426 (34.9)

568 (28.4)

<0.01

Vasopressor use in 24 h, n (%)

180 (14.7)

267 (13.4)

0.28

Vasopressor use in 48 h, n (%)

193 (15.8)

295 (14.8)

0.43

Max vasopressor doseb (Norepinephrine equivalents, mcg/kg/min)

0.30 (0.78)

0.26 (0.40)

0.51

Intubation within 48 h, n (%)

119 (9.7)

156 (7.8)

0.06

Total duration of antibiotics, days

4.8 (4.2)

5.2 (4.1)

0.01

Number of piperacillin-tazobactam doses

12.5 (10.6)

11.9 (10.3)

0.13

ED length of stay, hours

6.0 (5.5)

5.4 (3.5)

<0.01

Disposition, n (%)

<0.01

Intensive care unit

487 (39.8)

698 (34.9)

Operating room

56 (4.6)

216 (10.8)

Regular nursing floor

677 (55.4)

1076 (53.9)

Transfer/discharge from ED

2 (0.2)

7 (0.4)

Data are described as mean (SD) unless otherwise specified. ASE = adult sepsis event; ED = emergency department; SOFA = Sequential Organ Failure Assessment.

a Asian/Indian/Multinational/Other.

b Including only patients on vasopressors.

In the subgroup analysis of patients meeting ASE criteria, 28.6% of pa- tients in the pre-update group had a major delay to second dose piperacillin-tazobactam administration vs 23.6% of patients in the

Image of Fig. 1

Fig. 1. Flow diagram of patient identification and inclusion in the study.

Outcomes.

Outcome Measure

Pre-Update Group

Post-Update Group

p-value

Logistic Regression Analysis, aORc

(n = 1222)

(n = 1997)

(95% CI)

Major delaya to second dose piperacillin-tazobactam, n (%)

400 (32.7)

514 (25.7)

<0.01

0.66d (0.55 to 0.81)

Time to second dose piperacillin-tazobactam, hoursb

7.6 (6.1 to 9.9)

7.1 (5.8 to 8.7)

<0.01

In-hospital mortality, n (%)

85 (7.0)

120 (6.0)

0.29

1.02e (0.71 to 1.47)

Mechanical ventilation duration, days

5.6 (7.0)

4.2 (4.6)

0.02

Hospital length of stay, days

7.3 (5.8)

7.0 (6.3)

0.18

Data are described as mean (SD) unless otherwise specified. aOR = adjusted odds ratio; IQR = interquartile range; SOFA = Sequential Organ Failure Assessment; ASE = adult sepsis event; ED = emergency department.

a Major delay defined as delay >25% recommended dosing interval based on renal function.

b Expressed as median (IQR).

c aORs are derived by comparing post-update group relative to the pre-update group.

d Adjusted for baseline SOFA score, ASE criteria, initial ED lactate, second piperacillin-tazobactam dose due while in the ED, second piperacillin-tazobactam dose due within 1 h of transfer, age, gender, ED disposition, and hospital.

e Adjusted for baseline SOFA score, ASE criteria, Charlson Comorbidity Index, vasopressor use, mechanical ventilation, piperacillin-tazobactam administration within 1 h of lactate collection, and ED disposition.

post-update group (absolute difference -5% [95% CI -10.6% to 0.5%]) (Table 3). No differences were detected in in-hospital mortality or hos- pital length of stay between the pre- and post-update groups in this subgroup analysis, but duration of mechanical ventilation was shorter in the post-update group (6.8 [7.9] vs 4.8 [4.8] days, p = 0.03). Results for the subgroup analysis of patients with septic shock were similar to the ASE subgroup (Table 3). In the sensitivity analysis using an alterna- tive definition for major second dose delay, patients in the post-update group less frequently experienced major delay in administration of sec- ond piperacillin-tazobactam dose (32.5% vs 27.0%, absolute difference

-5.4% [95% CI -2.2% to -8.7%]). Further, after multivariable adjust- ment, order set design update was independently associated with lower odds of major delay (OR = 0.72, [95% CI 0.59 to 0.88]).

  1. Discussion

This study examined the association between sepsis order set up- dates with scheduled antibiotic frequencies and the proportion of pa- tients with major delay in second dose piperacillin-tazobactam administration. Before order set update, 32.7% of patients in the present study experienced a delay in second dose of piperacillin-tazobactam ad- ministration, similar to the rate of second dose antibiotic delay reported in prior evaluations examining this issue (33%) [2]. Overall, we found that updating an ED sepsis order set from one time antibiotic frequen- cies to scheduled frequencies was associated with 34% lower odds of experiencing a delay in administration of the second dose of piperacillin-tazobactam.

Image of Fig. 2

Fig. 2. Interrupted time series (segmented regression) analysis of delay in second dose of antibiotics.

Interrupted time series analysis of proportion of second dose delays over time. The dashed vertical line represents the point in time in which the ED sepsis order set was updated (inter- ruption point = 5/7/2019). Data points represent the monthly proportion of patients who experienced a delay in their second dose of antibiotics, data points on the left side of the inter- ruption point indicate the proportion of patients before the order set was updated, and data points on the right indicate proportion of patients after the order set was updated. Lines represent the regression lines for delays in second dose antibiotics during the study time period before and after the order set was updated. Results showed a pre-update slope of 0.29% (95% CI, -0.54 to 1.12%) and a post-update slope of 0.32% (95% CI, -0.24 to 0.67%) per month. There was no significant change in slope detected after the order set was updated (change in slope: 0.03% [95% CI, -0.87 to 0.93]). After the ED order set was updated, there was a significant level change in the proportion of patients who experienced a delay in their second dose of antibiotics: -10% (95% CI, -18 to -2%).

Table 3

Subgroup analyses.

Patients Meeting Adult Sepsis Event (ASE) Criteria (n = 994)

Outcome Measure

Pre-Update Group (n = 426)

Post-Update Group (n = 568)

p-value

Logistic Regression Analysis, aORc (95% CI)

Major delaya to second dose

122 (28.6)

134 (23.6)

0.07

0.81d (0.59 to 1.11)

piperacillin-tazobactam, n (%)

Time to second dose piperacillin-tazobactam, hoursb

7.7 (6.2 to 10.2)

7.2 (5.8 to 8.8)

< 0.01

In-hospital mortality, n (%)

44 (10.3)

44 (7.8)

0.16

0.77e (0.46 to 1.28)

Mechanical ventilation duration, days

6.8 (7.9)

4.8 (4.8)

0.03

Hospital length of stay, days

9.8 (6.5)

9.5 (7.2)

0.52

Patients with Septic Shockf (n = 447)

Outcome Measure

Pre-Update Group (n = 180)

Post-Update Group (n = 267)

p-value

Logistic Regression Analysis, aORc (95% CI)

Major delaya to second dose piperacillin-tazobactam, n (%)

45 (25.0)

64 (24.0)

0.80

0.90g (0.54 to 1.48)

Time to second dose piperacillin-tazobactam, hoursb

8.0 (6.4 to 11.1)

7.3 (5.6 to 10.0)

0.01

In-hospital mortality, n (%)

40 (22.2)

45 (16.9)

0.17

0.83h (0.46 to 1.51)

Mechanical ventilation duration, days

4.4 (3.8)

4.0 (3.8)

0.57

Hospital length of stay, days

9.6 (7.1)

9.2 (8.2)

0.62

Data are described as mean (SD) unless otherwise specified. aOR = adjusted odds ratio; IQR = interquartile range; SOFA = Sequential Organ Failure Assessment; ED = emergency department.

a Major delay defined as delay >25% recommended dosing interval based on renal function.

b Expressed as median (IQR).

c aORs are derived by comparing post-update group relative to the pre-update group.

d Adjusted for baseline SOFA score, initial ED lactate, second piperacillin-tazobactam dose due while in the ED, second piperacillin-tazobactam dose due within 1 h of transfer, age, gender, ED disposition, and hospital.

e Adjusted for baseline SOFA score, Charlson Comorbidity Index, vasopressor use, mechanical ventilation, piperacillin-tazobactam administration within 1 h of lactate collection, and ED disposition.

f Septic shock defined as patients on vasopressors.

g Adjusted for baseline SOFA score, ASE criteria, initial ED lactate, second piperacillin-tazobactam dose due while in the ED, second piperacillin-tazobactam dose due within 1 h of transfer, age, gender, ED disposition, and hospital.

h Adjusted for baseline SOFA score, Charlson Comorbidity Index, ASE criteria, mechanical ventilation, piperacillin-tazobactam administration within 1 h of lactate collection, and ED disposition.

To minimize delays, current guidelines prioritize timely first dose antibiotics for sepsis patients in the ED [1,21,22]. Multiple strategies for improving time to first dose antibiotics have been reported in the lit- erature including early screening, checklists, order sets/panels, staff ed- ucation, local stocking of antibiotics, and electronic medical record Clinical decision support [23-28]. Assessment of second dose antibiotic delay is increasingly being reported and has also been associated with adverse outcomes [2-9]. Only one other study examined a potential so- lution to reduce second dose antibiotic delays where pharmacists inter- vened on select antibiotic orders in the ED and scheduled subsequent doses of antibiotics for a 24-h period [10]. Scheduled antibiotic ordering by ED pharmacists significantly reduced the proportion of patients who experienced major delay to second dose and was associated with de- creased in-hospital mortality [10]. Although an extremely beneficial process improvement, this intervention was retroactive and required screening and recognition of high-risk ED patients by pharmacists to prevent subsequent delays following one-time antibiotic orders and uti- lized ED-based pharmacists not readily available 24/7 in all EDs [10]. Our analysis is the first to examine the impact of a proactive interven- tion at the time of initial antibiotic ordering which was implemented to prevent inadvertent delays.

Due to broad differential diagnoses of ED patients, ED clinicians com- monly utilize one time doses of antibiotics to complete patient workups and allow care to transition to admitting teams. However, one-time orders require subsequent prompt ED and/or inpatient clinician action to ensure continuity of therapy and prevent subsequent dose delays. Previous literature has demonstrated that major second dose delay may be associated with worse patient outcomes, including mortality [2,4,7-9]. The largest study published to date reported that delay in second dose antibiotic administration greater than one hour was not as- sociated with in-hospital mortality in their overall cohort of patients with suspected sepsis but was associated with significantly increased in-hospital mortality in their subgroup of patients with septic shock

(OR 1.34, [95% CI 1.05 to 1.70]) suggesting second dose delay is even more detrimental in patients who are more severely ill [9].

In this study, updating an ED sepsis order set design to include scheduled antibiotic frequencies was not associated with reduced in- hospital mortality or hospital length of stay. This finding may partially be explained by the relatively low severity of illness in our study sample as evidenced by low SOFA scores and low proportions of patients who received vasopressors (15.2%) or mechanical ventilation (8.5%) within 48 h of admission. Additionally, only 30.9% of study patients met ASE criteria for sepsis. Although more patients had piperacillin-tazobactam ordered through the ED sepsis order set in the post-update period, the proportion of patients meeting ASE criteria was significantly lower and the modest absolute reduction in delay (approximately one hour) may have contributed to the lack of an observed difference in mortality between groups in this potentially less critically ill cohort. While our study did not identify an association with mortality, inclusion of sched- uled antibiotic frequencies in our ED sepsis order set was associated with significantly reduced odds of major delay which could support a move to routine use of scheduled antibiotic frequencies in the ED.

In the subgroup of patients meeting ASE criteria, there were no dif- ferences observed in major delay or in-hospital mortality between pre- and post-update groups. Additionally, order set update was not as- sociated with either major delay or in-hospital mortality when adjusting for confounding factors in multivariable regression analyses. To further investigate the association between order set update and in-hospital mortality in the most critically ill patients in our study, we conducted a post-hoc subgroup analysis of patients with septic shock. Similar to the ASE subgroup analysis, order set update in the septic shock subgroup was not associated with in-hospital mortality (aOR 0.83; [95% CI 0.46 to 1.51]). These findings could be explained by a vari- ety of factors including higher prioritization of antibiotic administration in patients who appear sicker upon presentation as well as potential dif- ferences in fluid resuscitation, appropriateness of Empiric antibiotics,

source control, and ED volume which we were unable to control for in our analyses. Additionally, there were no significant differences in major delay between pre- and post-update groups, which likely con- tributed to the lack of an observed association between order set update and in-hospital mortality in these subgroups. Finally, our subgroup analyses were likely underpowered to evaluate these outcomes.

We conducted a sensitivity analysis using second piperacillin- tazobactam order frequency as the metric for assessing major delay in both groups rather than renal function assessment to define recom- mended frequency. Using actual ordered frequency in this sensitivity analysis minimizes the risk of detection bias and addresses the concern of inconsistency between recommended dosing interval based on renal function and the actual frequency ordered by a clinician. Importantly, the results of this sensitivity analysis corroborated and complemented the findings in our primary analysis.

Despite order set updates, approximately 25% of ED patients experi- enced major delay to second dose piperacillin-tazobactam administra- tion. Further, the proportion of patients in our study who received their first dose of piperacillin-tazobactam within one hour of first lactate collection was lower in the post-update group, and our interrupted time series analysis showed the proportion of patients experiencing second dose delay increased by 0.32% each month in the study period. As such, there is opportunity for additional quality improvement initiatives beyond order set design changes to minimize antibiotic delays, includ- ing use of implementation science methods to sustain initial improve- ments over time [29].

This study has many strengths including its large size, multicenter evaluation, pragmatic design, and a relatively simple intervention method. This study also addresses a critical clinical question of how to mitigate the risk of major second dose delay and has important implica- tions for ED sepsis care. Limitations of this study include its retrospec- tive nature with inherent risk of various biases and confounding. However, we attempted to control for factors most likely associated with our primary outcome by using logistic regression modeling. The study sample represents a diverse health-system population and there are likely practice differences between sites that could have influenced our results. However, hospital was included in our regression analysis to control for this effect. Additionally, while use of the ED sepsis order set is strongly encouraged and clinician usage is tracked as a quality metric, clinicians are not required to order antibiotics through this order set, and, consequently, additional sepsis and septic shock patients were likely omitted from our study. Also, because collecting the time of sepsis identification retrospectively is challenging, lactate collection time was used as a surrogate for this measure and may not accurately represent the time Clinical concern for sepsis occurred. Lastly, only piperacillin- tazobactam was studied because it is our institution’s primary anti- Pseudomonal beta-lactam and results may not be generalizable to other antibiotics with less frequent dosing intervals.

  1. Conclusions

Introduction of scheduled antibiotic frequencies in a standardized health-system ED sepsis order set was associated with a significant re- duction in major delay to second dose piperacillin-tazobactam adminis- tration. Designing ED sepsis order sets with scheduled antibiotic frequencies may be a pragmatic mechanism to decrease delays in sec- ond antibiotic doses. Future research should be focused on additional strategies to mitigate second dose antibiotic delay and associations with patient outcomes.

CRediT authorship contribution statement Reaghan M. Erickson: Writing – review & editing, Writing – original

draft, Validation, Project administration, Methodology, Investigation,

Formal analysis, Data curation, Conceptualization. Gretchen L. Sacha:

Writing – review & editing, Validation, Supervision, Methodology,

Formal analysis, Conceptualization. Seth R. Bauer: Writing – review & editing, Validation, Supervision, Methodology, Formal analysis, Concep- tualization. Baruch S. Fertel: Writing – review & editing, Methodology, Conceptualization. Matthew R. Dettmer: Writing – review & editing, Methodology, Conceptualization. Jessica L. Wesolek: Writing – review & editing, Methodology, Conceptualization. Matthew J. Campbell: Writing – review & editing, Writing – original draft, Validation, Supervision, Project administration, Methodology, Formal analysis, Conceptualization.

Declaration of Competing Interest

SRB is supported by a grant from the National Institutes of Health, National Institute of General Medical Sciences [grant number K08GM147806]. The funding sources had no role in study design; data collection, analysis, or interpretation; writing the report; or the decision to submit the report for publication. Its contents are solely the responsi- bility of the authors and do not necessarily represent the official views of the National Institutes of Health. The remaining authors have no con- flicts of interest.

References

  1. Kumar A, Roberts D, Wood KE, et al. Duration of hypotension before initiation of effective antimicrobial therapy is the critical determinant of survival in human septic shock. Crit Care Med. 2006;34(6):1589-96.
  2. Leisman D, Huang V, Zhou Q, et al. Delayed second dose antibiotics for patients admitted from the emergency department with Sepsis: prevalence, risk factors, and outcomes. Crit Care Med. 2017;45(6):956-65.
  3. Andersson M, Ostholm-Balkhed A, Fredrikson M, et al. Delay of appropriate antibi- otic treatment is associated with high mortality in patients with community-onset sepsis in a Swedish setting. Eur J Clin Microbiol Infect Dis. 2019;38(7):1223-34.
  4. Hammer J, Merchant S, Zhao J, et al. Impact of delays in second-dose beta lactam antibiotics on mortality in sepsis and septic shock. Crit Care Med. 2020;48(1):15.
  5. Randolph JL, Chan K, Albright A, Chen A. Delays in administration of the second antibiotic dose in patients with severe Sepsis and septic shock. Hosp Pharm. 2021; 56(4):247-51.
  6. Subedi B, Kim S, Nguyen N. Evaluation of second-dose antibiotic in patients with sepsis: multisite quality improvement study. Crit Care Med. 2019;47(1):775.
  7. Lykins VJD, Kuttab HI, Rourke EM, et al. The effect of delays in second-dose antibi- otics on patients with severe sepsis and septic shock. Am J Emerg Med. 2021;47: 80-5.
  8. Kemmler CB, Sangal RB, Rothenberg C, et al. Delays in antibiotic redosing: associa- tion with inpatient mortality and risk factors for delay. Am J Emerg Med. 2021;46: 63-9.
  9. Taylor SP, Shah M, Kowalkowski MA, Taylor B, Chou SH. First-to-second antibiotic delay and hospital mortality among emergency department patients with suspected sepsis. Am J Emerg Med. 2021;46:20-2.
  10. Payne-Cardona M, San Luis VA, Aazami R, et al. Pharmacist driven antibiotic redosing in the emergency department. Am J Emerg Med. 2021;50:160-6.
  11. Crass RL, Rodvold KA, Mueller BA, Pai MP. Renal dosing of antibiotics: are we jumping the gun? Clin Infect Dis. 2019;68(9):1596-602.
  12. Cleveland Clinic Foundation. Guidelines for Antimicrobial Usage, 2020-2021; 2020.
  13. Rhee C, Dantes R, Epstein L, et al. Incidence and trends of Sepsis in US hospitals using clinical vs Claims data, 2009-2014. JAMA. 2017;318(13):1241-9.
  14. Centers for Disease Control and Prevention: Hospital Toolkit for Adult Sepsis Surveil- lance. Available at: https://www.cdc.gov/sepsis/pdfs/Sepsis-Surveillance-Toolkit- Mar-2018_508.pdf; 2018.
  15. Churpek MM, Dumanian J, Dussault N, et al. Determining the electronic signature of infection in electronic health record data. Crit Care Med. 2021;49(7):e673-82.
  16. Lederer DJ, Bell SC, Branson RD, et al. Control of confounding and reporting of results in causal inference studies. Guidance for authors from editors of respiratory, sleep, and critical care journals [published correction appears in Ann am Thorac Soc. 2019 Feb;16(2):283]. Ann Am Thorac Soc. 2019;16(1):22-8.
  17. Linden A. A matching framework to improve causal inference in interrupted time- series analysis. J Eval Clin Pract. 2018;24(2):408-15.
  18. Kontopantelis E, Doran T, Springate DA, Buchan I, Reeves D. Regression based quasi- experimental approach when randomisation is not an option: interrupted time series analysis. BMJ. 2015;350:h2750.
  19. Wagner AK, Soumerai SB, Zhang F, Ross-Degnan D. Segmented regression analysis of interrupted time series studies in medication use research. J Clin Pharm Ther. 2002; 27(4):299-309.
  20. Jandoc R, Burden AM, Mamdani M, et al. Interrupted time series analysis in drug uti- lization research is increasing: systematic review and recommendations. J Clin Epidemiol. 2015;68(8):950-6.
  21. Seymour CW, Gesten F, Prescott HC, et al. Time to treatment and mortality during mandated emergency Care for Sepsis. N Engl J Med. 2017;376(23):2235-44.
  22. 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(11):e1063-143.
  23. Emerson BL, Ciaburri R, Brophy C, Kandil SB. Decreasing time to antibiotics for pa- tients with Sepsis in the emergency department. Pediatr Qual Saf. 2019;4(3):e173.
  24. Emerson BL, Prozora S, Jacob A, et al. An initiative to decrease time to antibiotics for patients with fever and neutropenia. Am J Med Qual. 2019;34(2):158-64.
  25. Hunter CL, Silvestri S, Stone A, et al. Prehospital sepsis alert notification decreases time to initiation of CMS sepsis core measures. Am J Emerg Med. 2019;37(1):114-7.
  26. Mittal Y, Sankar J, Dhochak N, Gupta S, Lodha R, Kabra SK. Decreasing the time to Ad- ministration of First Dose of antibiotics in children with severe Sepsis. J Healthc Qual. 2019;41(1):32-8.
  27. Mitzkewich M. sepsis screening in triage to decrease door-to-antibiotic time. J Emerg Nurs. 2019;45(3):254-6.
  28. Moore WR, Vermuelen A, Taylor R, Kihara D, Wahome E. Improving 3-hour Sepsis bundled care outcomes: implementation of a nurse-driven sepsis protocol in the emergency department. J Emerg Nurs. 2019;45(6):690-8.
  29. Weiss CH, Krishnan JA, Au DH, et al. An official American Thoracic Society research statement: implementation science in pulmonary, critical care, and sleep medicine. Am J Respir Crit Care Med. 2016;194(8):1015-25.

Leave a Reply

Your email address will not be published. Required fields are marked *