a b s t r a c t

Purpose: Optimal sepsis outcomes are achieved when sepsis is recognized early. Recognizing sepsis in the prehos- pital, EMS setting can be challenging and unreliable. The purpose of this study is to evaluate whether implemen- tation of an EMS sepsis screening and prehospital alert protocol called PRESS (PREhospital SepsiS) is associated with improved sepsis recognition by EMS providers.

Design: We conducted a 12-month, before-after implementation study of the PRESS protocol in a large, public EMS system. The study intervention was a PRESS training program delivered to EMS providers. EMS patient in- clusion criteria included: age >= 18 years, EMS systolic blood pressure < 110 mmHg, EMS heart rate > 90 bpm, and EMS respiratory rate > 20 bpm. Study exclusion criteria included the presence of any of following EMS condi- tions: trauma, cardiac arrest, pregnancy, toxic ingestion, or psychiatric emergency. Retrospective chart review was performed on all eligible EMS encounters during the study period. The primary outcome variable was the proportion of patients with sepsis who were identified by EMS providers.

Results: Approximately 300 EMS providers were trained to use PRESS. A total of 498 patient encounters met criteria for study inclusion; 222 were excluded, primarily due to trauma. A total of 276 patient encounters were analyzed. Sepsis recognition by EMS providers increased from 12% pre-PRESS protocol to 59% post-PRESS protocol (p < 0.001). In a post-hoc analysis of the post-PRESS cohort, septic patients who were identified by EMS received antibiotics 24 min faster than septic patients who were not identified by EMS [28 min (IQR 18-48) vs 52 (IQR 27-98), respectively, p = 0.021].

Conclusion: Implementation of an EMS sepsis screening and prehospital alert protocol was associated with an in- crease in sepsis recognition rates by EMS providers and a decrease in time to first antibiotic administration in the emergency department. Further studies are needed to evaluate the impact of this protocol in other populations.

(C) 2022

1. Introduction

Sepsis is a life-threatening, time-sensitive medical emergency en- countered frequently in the prehospital setting [1]. Community-onset sepsis accounts for more than 80% of sepsis hospitalizations, and EMS is the point of first medical contact in an estimated 40% of these cases [1,2]. Despite the high incidence of prehospital sepsis, EMS providers largely lack the foundational tools and protocols necessary to facilitate

* Corresponding author at: Division of Pulmonary, Allergy, Critical Care, and Sleep Medicine, Emory University School of Medicine, 615 Michael Street, Suite 205, Atlanta, GA 30322, United States of America.

E-mail address: [email protected] (C.C. Polito).

reliable recognition in ambulances. Notably, it is estimated that only 10-20% of sepsis cases are identified by EMS providers using clinical judgement alone [3,4].

Developing reliable and effective Screening tools for sepsis, analo- gous to a screening EKG for STEMI or a Neurologic assessment for stroke, is one important step toward improving prehospital sepsis recognition rates and care. Ultimately, a comprehensive prehospital sepsis pathway, similar to well-established EMS care pathways for other life-threatening Medical emergencies including STEMI and stroke, may be helpful to le- verage opportunities at the point of first medical contact that translate into better patient outcomes at hospital discharge.

The purpose of this study is to evaluate the impact of the PRESS (PREhospital SepsiS) screening and prehospital protocol on EMS pro- vider sepsis recognition rates [3].

https://doi.org/10.1016/j.ajem.2022.04.035

0735-6757/(C) 2022

1. Methods
   1. Design, setting and patient selection

We conducted a 12-month, before-after study of PRESS protocol im- plementation in a large, public EMS system in Atlanta, Georgia between February 2016 and January 2017 [3]. The pre-post study design is pre- sented in Fig. 1. Baseline data was collected prior to PRESS implementa- tion for the first 6 months of these study. This data was used as comparison group data. Immediately prior the intervention phase, all EMS providers (approximately 300) were trained to use the PRESS pro- tocol displayed in Fig. 2. PRESS is comprised of 2 major parts: 1) an evidence-based, EMS sepsis screening tool, and 2) a prehospital sepsis alert called via 2-way radio to the receiving hospital for all patients who screened positive for sepsis. PRESS protocol training occurred dur- ing a two-week period in August 2016. Patient data was not collected during the training period in order to minimize the risk of overlap be- tween groups. All eligible EMS patients included in study analysis were transported to a large, public hospital in downtown Atlanta.

Study inclusion criteria included all of the following EMS criteria:

1) age 18 years or greater, 2) EMS systolic blood pressure < 110 mmHg, 3) EMS heartrate >90 beats per minute, and 4) EMS respiratory rate > 20 breaths per minute. Patients were excluded for any of the fol- lowing EMS conditions: 1) trauma, 2) cardiac arrest, 3) pregnancy,

4) toxic ingestion, 5) psychiatric emergency. Patients were also ex- cluded for inability to link EMS and hospital patient records, or if the pa- tient left the emergency department before being evaluated by a medical provider. Both of these conditions, when present, precluded the ability of the study team to categorize the presence or absence of sepsis.

Manual review and linkage of EMS and hospital records was per- formed to identify eligible patients and collect study data elements. Sep- sis was defined using Sep-2 criteria which were in use at the time of study onset (February 2016). Patients were categorized as having sepsis or not based on chart review of the first 48 h of patient hospital data [5]. The primary outcome measure was the proportion of EMS patients with sepsis identified by EMS providers. Patients in before and after groups were compared. Secondary process outcomes were pre- specified and included elements of the 3-h Sepsis Bundle administered in the emergency department: 1) blood culture order, 2) lactic acid order, 3) IV fluid administration, and 4) antibiotic administration. Sec- ondary patient outcomes were also pre-specified and included:

1) need for ICU admission, 2) need for invasive mechanical ventilation,

3) hospital length of stay, and 4) in-hospital mortality.

• 1. PRESS protocol

The EMS-based PRESS protocol is presented in Fig. 1 and includes the following steps: 1) identify patients eligible for screening, 2) perform PRESS screening on all eligible patients, 3) call a prehospital alert to the receiving hospital for all positive sepsis screens, and 4) document a positive sepsis screen and prehospital alert in the EMS electronic health record. During the baseline phase of data collection, EMS did not use another prehospital sepsis screening tool or a standardized pre- hospital sepsis alert.

Fig. 2. PRESS Protocol.

• 1. EMS training curriculum

Approximately 300 EMS providers were trained during a two- week training period in August 2016. Patient data was not collected during the 2-week training period to avoid crossover between study groups. The study and EMS leadership teams collaborated to develop and implement a multifaceted sepsis training toolkit over a variety of educational platforms. The PRESS training toolkit included: 1) a mandatory one hour in-person, in-service training session, 3) a computer-based learning (CBL) module followed by a 10-question post-test, 3) daily shift huddle reminders by shift leaders and field training officers, 4) distribution of laminated protocol badge buddies to all EMS providers, and 5) posting of laminated protocols inside all ambulances. The study protocol was approved by the Emory IRB and Grady Research Oversight Committee with waiver of informed consent.

Baseline phase (6 months)	2 weeks	Intervention phase (6 months)
Pre-PRESS protocol	PRESS training	Post-PRESS protocol
Baseline data collection	No data collection	Intervention phase data collection

Fig. 1. Pre-post study design.

498 EMS patients screened

222 (45%) EMS patients excluded

87 trauma

43 psychiatric emergency

12 unable to link medical record data 9 cardiac arrest

9 pregnant patient

9 patient left ED before being seen by provider

53 other

Fig. 4. EMS sepsis recognition rates before and after PRESS implementation.

276 EMS patients included for study analysis

• 1. Analysis

Fig. 3. Patient flow diagram.

ethnicity, and severity of acute illness as indicated by the APACHE II score. There was a higher proportion of patients with diabetes mellitus in the post-PRESS group compared to pre-PRESS (18% vs 3%, respec- tively, p < 0.041).

The primary outcome measure for this study, the proportion of sep-

Counts (n) and percentages are presented for all categorical vari- ables, along with p-values from Chi-Square tests. Means and standard deviations are presented for continuous variables, along with p-values from two-sample independent t-tests. To calculate APACHE II score, the following missing values were assigned zero points: 1) blood gas pH, 2) FiO₂. Three-hour bundle compliance rates were analyzed with Fisher’s exact tests and Wilcoxon Rank Sum tests due to non-normal distribution of a continuous variable. All statistical tests were conducted in SAS 9.4 with statistical significance level of 0.05.

1. Results

A total of 498 EMS patients met study inclusion criteria during the 12-month study period (Fig. 3). Of these, 222 (45%) were excluded, with trauma being the most common reason for exclusion (n = 87). A total of 276 (55%) EMS patients were included for study analysis. Of these 276 patients, 130 (47%) were categorized as having sepsis.

Characteristics of EMS patients with sepsis are shown in Table 1. Pre and post-PRESS groups were similar with respect to age, sex, race/

tic patients identified by EMS personnel, was significantly higher post- PRESS compared to pre-PRESS (59% vs 12%, respectively p < 0.001; Table 1). A run chart, or monthly longitudinal display, of sepsis recogni- tion rates over the course of the 12-month study is shown in Fig. 4.

Patients with sepsis had similar EMS vital signs and received similar interventions during the EMS phase of care before and after PRESS im- plementation (Table 2). Emergency department care including 3-h sep- sis bundle compliance was the unchanged or higher after PRESS training (Table 3). Table 4 shows a post-hoc analysis of the post-PRESS group, stratified by whether EMS providers identified the patient as septic using the PRESS protocol or not. Patients with sepsis who were identi- fied using the PRESS protocol received blood culture and antibiotic or- ders in the emergency room more often than patients who were not identified by PRESS (84% pre-PRESS and 100% post-PRESS for both blood cultures and antibiotics, p = 0.005; Table 4).

Additionally, patients with sepsis who were identified by EMS per- sonnel received antibiotics in the emergency department 28 min faster compared to patients not identified as septic by EMS providers [median

Table 1

Table 2

EMS characteristics of patients with sepsis before and after PRESS implementation.

Characteristics of patients wit	h sepsis before and afte	r PRESS implementation.		Characteristic	Pre-PRESS	Post-PRESS	p-value
Characteristic	Pre-PRESS (n = 51)	Post-PRESS (n = 79) p-value			(n = 51)	(n = 79)
Age (years)	57 (17)	59 (16) 0.404		Systolic blood pressure (mmHg)	89 (15)	90 (12)	0.924
Sex				Heart rate (bpm)	128 (21)	126 (20)	0.589
Women	24 (47)	37 (47) 0.980		Respiratory rate (bpm)	29 (8)	29 (9)	0.903
Men	27 (53)	42 (53)		Oxygen saturation (%)	92 (7)	92 (6)	0.510
Race				Blood glucose (mg/dL)	162 (99)	171 (121)	0.681
White	5 (10)	12 (15) 0.436		Glascow Coma Scale	12 (4)	12 (4)	0.672
Black	44 (86)	66 (84)		IV attempted or placed	42 (82)	67 (85)	0.710
Other	2 (4)	1 (1)		IV fluid administered	27 (53)	44 (56)	0.758
Comorbid Conditions				Volume of intravenous fluid administered	555 (377)	410 (264)	0.061
Hypertension	26 (51)	44 (56) 0.598		(mL)
Diabetes mellitus	17 (3)	14 (18) 0.041		EMS dispatch to first medical contact time	11 (5)	12 (6)	0.227
Congestive heart failure	6 (12)	13 (16) 0.460		(min)
Chronic lung disease	12 (24)	15 (19) 0.533		First medical contact to hospital arrival time	38 (9)	38 (12)	0.750
end stage renal disease	1 (2)	5 (6) 0.246		(min)
APACHE II score	19 (10)	19 (8) 0.843		Total EMS encounter time (min)	49 (10)	51 (13)	0.403
Sepsis identified by EMS 6 (12) 47 (59) <0.001 * N (%) and p-value calculated from chi-square test are reported for IV attempted or placed

* Mean (SD) and p-value calculated from two sample independent t-test are reported for Age and APACHE II score. N (%) and p-value calculated from chi-square test are reported for Sex, Race, Comorbid Conditions, and Sepsis identified by EMS to compare Pre and Post.

(yes) and IV fluid administered (yes). For the remaining characteristics, mean (SD) and p- value calculated from two sample independent t-test are reported to compare Pre and Post.

Table 3

Emergency department care including 3-h sepsis bundle compliance before and after

PRESS implementation.
Characteristic	Pre-PRESS	Post-PRESS	p-value
	(n = 51)	(n = 79)
Intravenous fluid ordered	44 (86)	70 (89)	0.693
Lactic acid level ordered	47 (92)	79 (100)	0.011
Blood cultures ordered	46 (90)	74 (94)	0.468
Antibiotics ordered	46 (90)	74 (94)	0.468
Time to intravenous fluid administration (min)	27 (17,46)	24 (18,46)	0.995
Time to lactic acid level order (min)	28 (17,45)	23 (17,40)	0.628
Time to blood culture order (min)	36 (19,59)	23 (17,42)	0.096
Time to antibiotic administered (min)	46	37 (20,71)	0.226
	(21,100)

* Median (Q1, Q3) or n (%) are reported. The non-parametric p-value is calculated by the Wilcoxon Rank Sum test for continuous covariates and Fisher’s exact test for categorical covariates.

24 min (IQR 18-48) vs 52 (IQR 27-98), respectively, p = 0.021]. The proportion of patients admitted to the ICU, hospital length of stay, and in-hospital mortality was similar in both groups (Table 5).

1. Discussion

A protocolized approach to sepsis in the EMS setting represents a paradigm shift that mimics the approach for other life-threatening, time-sensitive medical conditions including ST-elevate myocardial in- farction and stroke. These other conditions are associated with mature EMS Care pathways that start at the point of first medical contact in the EMS setting and facilitate a streamlined care pathway into the hos- pital setting to facilitate near-immediate delivery of definitive, life- saving treatment. Arguably, the development of a prehospital pathway would be of similar benefit to patients with sepsis. The PRESS protocol utilized in this study pairs a validated sepsis screening tool with a pre- hospital sepsis alert to the receiving hospital. The aims of this protocol are to: 1) increase sepsis recognition by EMS providers, and 2) commu- nicate possible sepsis cases to the emergency department without delay by using a prehospital sepsis alert.

To our knowledge, this is the first study to demonstrate an improve- ment in sepsis recognition by EMS providers after implementation of an EMS-based sepsis protocol. While the screening tool included in the protocol was initially validated retrospectively using a Mathematical model derived from an EMS cohort, this is the first published report of real-world effectiveness of PRESS [3]. In this single-center, before-after EMS study, PRESS protocol implementation was associated with an al- most 5-fold increase in prehospital sepsis recognition rates. PRESS im- plementation was also associated with a higher proportion of septic patients receiving all components of the 3-h sepsis bundle in the emer- gency department when patients were identified by EMS personnel. In addition, and potentially most impactful, post-PRESS patients with sep- sis who were EMS-identified received antibiotics 28 min faster as com- pared to patients who were not EMS-identified. Importantly, it has been shown that for every hour delay in antibiotic administration, sepsis mortality increased by 8% per hour [6]. Improvements in EMS sepsis rec- ognition rates and 3-h bundle compliance did not translate into im- provement in patient outcomes.

Although patient outcomes were similar for septic patients who were EMS-identified versus not, outcomes in sepsis patients are subject to influences well beyond what occurs in the ambulance and early hos- pital phases of care. Additionally, this study was not powered to detect a difference in patient outcomes. Rather, the primary purpose was to study how PRESS impacted sepsis recognition rates by EMS providers. Therefore, a lack of improvement in outcomes should not detract from the importance of the observed improvements in sepsis recognition rates by EMS personnel.

• 1. Limitations

There are limitations to this study. This is a single-center, Pre-post trial and therefore subject to concerns about unmeasured confounders and generalizability. It is our hope that results from this study provide useful preliminary information to support a multicenter, stepped- wedge design in the future. Another limitation is the Sep-2 definition that was utilized to define sepsis, which has since been replaced by the Sep-3 definition. The impact of this screening protocol may vary among patients who are confirmed to have acute organ dysfunction

Table 4

Emergency department care after PRESS implementation, stratified by EMS sepsis recognition status.

Characteristic	Post-PRESS			p-value
	Overall (n = 79)	EMS unidentified (n = 32)	EMS identified (n = 47)
Intravenous fluid ordered	70 (89)	26 (81)	44 (94)	0.089
Lactic acid level ordered	79 (100)	32 (100)	47 (100)	-
Blood cultures ordered	74 (94)	27 (84)	47 (100)	0.005
Antibiotics ordered	74 (94)	27 (84)	47 (100)	0.005
Time to intravenous fluid administration (min)	24 (18,46)	24 (16,45)	24 (19,47)	0.881
Time to lactic acid level order (min)	23 (17,40)	27 (17.5,38.5)	22 (16,42)	0.569
Time to blood culture order (min)	23 (17,42)	25 (16,45)	22 (17,42)	0.152
Time to antibiotic administered (min)	37 (20,71)	52 (27,98)	24 (18,48)	0.021

* For the ordered characteristics, N(%) of yes ordered are reported along with the p-values from chi-square test. For the continuoUS time to characteristics, median (Q1, Q3) and p-value from the Wilcoxon Rank Sum test are reported due to non-normal distributions.

Table 5

Inpatient care and outcomes among patients with sepsis before and after PRESS implementation.

	Pre-PRESS (n = 51)	Post-PRESS (n = 79)	p-value
Admitted to ICU	25 (49)	37 (47)	0.808
Days of mechanical ventilation	8 (7)	6 (4)	0.170
Hospital length of stay** (days)	9 (4,20)	8 (4,14)	0.259
In-hospital mortality	9 (18)	11 (15)	0.618

* N (%) are reported and Chi-square p-value are reported for Admitted to ICU and in-hospital mortality. Mean(SD) and the two sample independent t-test were utilized for days of mechanical ventilation. Due to violation of normality distribution, Median (Q1, Q3) and p-value from the Wilcoxon Rank Sum test were reported for hospital length of stay.

(e.g. Sep-3 definition) as compared to SIRS-based criteria. This should be evaluated in future study protocols.

In conclusion, implementation of the PRESS EMS sepsis screening protocol was associated with an increase in prehospital sepsis recogni- tion rates, increased compliance with the 3-h sepsis bundle, and de- creased time to first antibiotic administration in the emergency department. Further studies are needed to evaluate the impact of this protocol in other populations.

Funding acknowledgement

CP was supported by an Emory Medical Care Foundation grant to conduct this study.

Publication disclosure

Preliminary abstract data for this study was presented at the American Thoracic Society meeting in 2018. This manuscript has not been previously published or posted to any other journals and is not currently under review with any other journals.

Conflicts of interest declaration

All authors report having no conflicts of interest related to the con- duct of this study or the composition of this manuscript.

Credit authorship contribution statement

Carmen C. Polito: Writing - original draft, Supervision, Resources, Project administration, Methodology, Investigation, Funding acquisi- tion, Data curation, Conceptualization. Ingrid Bloom: Data curation, Writing - review & editing, Conceptualization. Charity Dunn: Writing - review & editing, Conceptualization. Renee’ Moore: Data curation, Formal analysis, Writing - review & editing. Jiandong Chen: Writing - review & editing, Formal analysis, Data curation. Arthur H. Yancey: Conceptualization, Project administration, Resources, Supervision, Writing - review & editing. Jonathan E. Sevransky: Writing - review & editing, Supervision, Project administration, Methodology, Investiga- tion, Conceptualization.

References

1. Seymour CW, et al. Severe sepsis in Pre-hospital emergency care: analysis of inci- dence, care, and outcome. Am J Respir Crit Care Med. 2012;186(12):1264-71.
2. Rhee C, et al. Epidemiology of hospital-onset versus community-onset sepsis in U.S. hospitals and association with mortality: a retrospective analysis using electronic clinical data. Crit Care Med. 2019;47(9):1169-76.
3. Polito CC, et al. Prehospital recognition of severe sepsis: development and validation of a novel EMS screening tool. Am J Emerg Med. 2015;33(9):1119-25.
4. Wallgren UM, et al. Identification of adult septic patients in the prehospital setting: a comparison of two screening tools and clinical judgment. Eur J Emerg Med. 2014;21 (4):260-5.
5. Rhodes A, et al. Surviving sepsis campaign: international guidelines for management of sepsis and septic shock: 2016. Crit Care Med. 2017;45(3):486-552.
6. Kumar A, 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.