a b s t r a c t

Background: It has not been investigated whether the quick Sepsis-related Organ Failure Assessment score (qSOFA), a new bedside tool for early sepsis detection, may help accelerating antibiotic initiation in ED patients with sepsis.

Methods: In this prospective pre/post quasi-experimental single-ED study, patients admitted with a suspected bacterial infection were managed using standard triage procedures only (baseline) or in association with qSOFA (intervention, with prioritization of patients with a qSOFA >= 2).

Results: A total of 151/328 (46.0%) and 185/350 (52.8%) patients with definite bacterial infection met the criteria for sepsis in the baseline and intervention periods, respectively. The sensitivity and specificity of a qSOFA >= 2 for sepsis prediction were 17.3% (95% confidence interval [CI], 13.6%-21.7%) and 98.8% (95% CI, 97.0%-99.5%). Eleven (7.3%) and 28 (13.5%) patients with sepsis in the baseline and intervention periods received a first antibiotic dose within one hour following triage (primary endpoint, absolute difference 6.2%, 95% CI [-0.5%, 12.7%], P = 0.08). The proportions of patients with sepsis receiving a first antibiotic dose within three hours following triage (39.7% [50/151] versus 36.8% [68/185], absolute difference - 2.9%, 95% CI [-13.3%, 7.3%], P = 0.65), requiring ICU ad- mission, or dying in the hospital were similar in both periods. The median ED occupation rate at triage was 104.3% (interquartile range [IQR], 80.4%-128.3%), with a median number of 157 ED visits per day (IQR, 147-169). Conclusions: A qSOFA-based triage procedure does not improve antibiotic timing and outcomes in patients with sepsis admitted to a high-volume ED. The qSOFA value at triage was poorly sensitive for early sepsis detection. Trial registration (ClinicalTrials.gov): NCT03299894.

(C) 2019

Introduction

* Corresponding author at: Medecine Intensive et Reanimation, Hopital de la Source, Centre Hospitalier Regional d’Orleans, 14, Boulevard de l’Hopital, 45100 Orleans, France.

E-mail address: [email protected] (F. Barbier).

Sepsis is defined by the occurrence of life-threatening organ dys- functions resulting from an unbalanced host response to infection [1]. The global incidence of sepsis is estimated as ranging from 15 to 30 mil- lion cases per year [2,3], with a 10% to 20% in-hospital fatality rate and poor long-term outcomes including impaired functional status,

https://doi.org/10.1016/j.ajem.2019.05.022

0735-6757/(C) 2019

accelerated cognitive decline, and late excess mortality [4-6]. These fea- tures, along with a substantial cost burden [7], make the improvement of sepsis management a pivotal public health priority.

Early recognition of sepsis is fundamental to hasten the implementa- tion of healthcare procedures that positively impact patient outcomes [8]. Although mixed [9,10], the available evidence supports a benefit of prompt antibiotic administration on survival rates in sepsis and septic shock, irrespective of the number of organ dysfunctions [11-13]. In ad- dition, the delayed initiation of adequate antibiotic therapy has been as- sociated with the worsening of baseline organ failures, the development of new failures, and extended lengths of hospital stay (LOS) [14]. Hence, the Surviving Sepsis Campaign (SCC) guidelines recommend that ad- ministration of Intravenous antibiotics be initiated as soon as possible and ideally within one hour after recognition of sepsis or septic shock [15].

Most of patients with community-acquired or healthcare-associated sepsis are admitted to the hospital through the emergency department (ED) [13]. Yet, the proportion of ED patients with sepsis receiving a first dose of antibiotic within one hour following triage remains consistently low as a combined effect of frequent atypical presentation, patient over- flow and organizational pitfalls that may all preclude a timely clinical evaluation by ED physicians [11,16-18]. In 2016, the Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3) elaborated the quick organ failure assessment score (qSOFA), a new bedside tool for early recognition of infected patients at-risk for sepsis-related in- hospital death or intensive care unit admission [19]. While its prognostic value has been the focus of several external validation stud- ies [20-26], it has not been investigated whether qSOFA may help accel- erating the initiation of antibiotics in patients with sepsis, especially in the ED.

The qSOFAST study was designed to test the hypothesis that routine

qSOFA calculation at triage may reduce the time to the first antibiotic administration in ED patients admitted with a bacterial infection and meeting the criteria for sepsis or septic shock.

Methods

Design and setting

This prospective, interventional, pre/post quasi-experimental study was conducted between October 2017 and June 2018 in the ED of a 1200-bed teaching hospital in France (60,955 ED visits of adult patients in 2017). During the first period (baseline), patients presenting with clinical signs suggestive of bacterial infection were stratified by Triage nurses and managed according to standard procedures, with tagging in the ED flow through a dedicated report form and the protocolized col- lection of blood and microbiological samples at triage before full medi- cal evaluation (Appendix). After completion of this first inclusion period, an educational program involving all ED physicians and nurses was implemented to introduce the routine calculation of qSOFA at triage (1 point for systolic blood pressure [SBP] <= 100 mmHg, respiratory rate

>= 22 per minute and altered mentation as defined by a Glasgow coma

scale [GCS] <= 13, with a total score ranging from 0 to 3) and consisted in formal courses, training sessions, poster campaign, and individual- ized feed-back by members of the study board. During the second pe- riod (intervention), the qSOFA score was calculated by triage nurses for every patient admitted with a suspected bacterial infection, in addi- tion to standard triage evaluation. Patients with a qSOFA value >= 2 were tagged and immediately referred to the attending senior ED physician for clinical evaluation and, when deemed appropriate, the initiation of empirical antibiotic therapy and other resuscitation measures. During both periods, nurses were asked to indicate in patient charts the exact timing (hh:mm) of each antibiotic administration. The study protocol was approved by the CPP Marseille/Sud-Mediterranee I ethic committee on June 8, 2017 (IRB no. 2017-A01000-53) and deposited on www. clinicaltrial.gov on October 3, 2017 (registration no. NCT03299894).

Results are reported according to the TREND guidelines (available at www.cdc.gov).

Selection of participants and data collection

All adult patients admitted with a suspected bacterial infection were eligible for inclusion (Appendix). Non-inclusion criteria were refusal of participation from the patient or his/her proxies, prior study enroll- ment, pregnancy, breast-feeding, imminent death, lack of coverage by the French public healthcare insurance system, and the administration of a first antibiotic dose prior to ED admission for patients transferred by a prehospital emergency medicalized unit.

Three investigators (J.P., J.P., and F.B.) reviewed all patient files re- lated to the ED visit and hospital stay and adjudicated consensually whether the ED admission was related to a bacterial infection based on the analysis of clinical features, biological and microbiological results, and radiological data. Investigator blinding for inclusion period was not possible given the pre/post design of the study. Patients in whom a def- inite diagnosis of bacterial infection was not retained were excluded from analyses.

For patients included during the Baseline period, the qSOFA items were retrospectively extracted from triage files for calculation of the score. For those included during the intervention period, the qSOFA value was recorded as calculated by triage nurses. Variables exposed in the tables were prospectively collected from electronic medical charts using a standardized case report form. Patients with sepsis or septic shock were identified through application of the Sepsis-3 defini- tions (Appendix); both subgroups were pooled for analyses and are hereafter referred to as “patients with sepsis”.

The ED occupation rate was calculated for every included patients as the ratio of the total number of patients present in the ED at the time of triage (that is, admitted and waiting for medical evaluation, under med- ical evaluation and management, or waiting for ward transfer)/the number of examination rooms in the ED (that is, 46).

Study endpoints

Pre-specified study endpoints were focused on patients who met the criteria for sepsis between triage (Day 0) and Day 1. The primary end- point was the administration of a first antibiotic dose in the ED within one hour following triage. Secondary endpoints included (i) the admin- istration of a first antibiotic dose within 3 h following triage, (ii) the ad- ministration of a first dose of adequate antibiotic within one hour and three hours following triage in patients with microbiologically docu- mented infection, (iii) a decrease in SOFA score >= 1 point between triage and Day 2, (iv) the need for ICU admission between triage and Day 2,

(v) overall and Day 7 in-hospital mortality, and (vi) hospital LOS.

Statistical analyses

We hypothesized that 40% of patients with sepsis during the base- line period and 60% of patients with sepsis during the intervention pe- riod would receive a first antibiotic dose within one hour following triage. Therefore, to demonstrate a significant difference on the primary endpoint (? set at 5%, power set at 80%, two-sided test), the inclusion of 97 patients with sepsis was required for each period. With the assump- tion that at least 25% of patients admitted with a suspected bacterial in- fection would present criteria for sepsis [11,25], a total of 780 patients (390 in each period) had to be included.

Categorical variables are expressed as number (percentage) and continuous variables as mean +- standard deviations (SD) or median (25th-75th percentile, IQR). Missing values for qSOFA items were as- sumed to be in the normal range. Missing values for SOFA items were assumed to be in the normal range at Day 1 and unchanged at Day 2. Pa- tient characteristics and study endpoints were compared using the ?2 test or the Fisher exact test for categorical variables and the Student’s

t-test or the Mann-Whitney test for continuous variables, as appropri- ate. The cumulative proportions of patients receiving a first antibiotic dose in the ED were compared using Kaplan-Meier curves and the log rank test. The performances of qSOFA to predict sepsis and in-hospital mortality were appraised by calculating sensitivity, specificity, and pos- itive and negative predictive values for a score >= 2, with construction of corresponding receiver operating characteristic curves (exploratory endpoints). All analyses were 2-tailed, and a P-value b0.05 was required for statistical significance. Analyses were performed using the R soft- ware version 3.5.1 (http://www.R-project.org) and the MedCalc soft- ware version 18 (Ostend, Belgium).

Results

Study population

Among the 780 enrolled patients (390 in each period), 328 in the baseline period (no qSOFA at triage - retrospective calculation) and 350 in the intervention period (routine qSOFA calculation at triage) had a definite bacterial infection and were retained for analyses (Fig. 1). The ED occupation rate at triage (117.4% [86.9%-136.9%] versus 91.3% [71.7%-113.1%], P b 0.0001) and the number of ED visits per day

(158 [148-172] versus 154 [145-162], P = 0.005) were slightly higher in the intervention period than in the baseline period.

Demographics and chronic diseases were evenly distributed (Table 1 and Table S1). Patients included in the intervention period presented more frequently with a respiratory rate >= 22 per minute while no vari- ation was observed for the SBP and GCS items of qSOFA (missing values are detailed in the Appendix). Twenty-one patients (6.5%) in the base- line period and 41 patients (11.7%) in the intervention period had a qSOFA >= 2 at triage (P = 0.02) (Table 1).

lower respiratory tract infections were more common in the inter- vention period (145 [41.4%] versus 108 [32.9%] patients, P = 0.03) while skin and soft tissue infections were more frequent in the baseline period (48 [14.6%] versus 31 [9.4%] patients, P = 0.02). Other Infection sites were well balanced (Table 1). Microbiological documentation was obtained in 40% of cases (Table S2). Overall, one or more antibiotic dose was administered in the ED in 241 (73.5%) and 267 (76.3%) pa- tients in the baseline and intervention periods, respectively (P = 0.42). Antibiotic classes are detailed in Table S1.

At Day 1, 151 patients (46.0%) in the baseline period and 185 patients (52.8%) in the intervention period met the criteria for sepsis (P = 0.08), including 7 (2.1%) and 3 (0.8%) patients with septic shock (P = 0.21). Organ failures not screened by the qSOFA score (that is, he- matological, hepatic and renal dysfunctions) occurred each in one third

Baseline period

No qSOFA calculation at ED triage

475 patients admitted with a suspected bacterial infection and screened for eligibility

22 patients meeting non-inclusion criteria Age <18 years: 13 Moribund at triage: 2

Pregnancy or breast-feeding: 5 In-ambulance antibiotic infusion: 1

62 excluded patients

No bacterial infection: 49 Patient refusal after emergency inclusion procedure: 9

Re-inclusion (2nd ED admission): 3 Pregnancy diagnosis after inclusion: 1

151 patients with sepsis and assessed for study endpoints

328 patients with definite bacterial infection

390 included patients

63 patients with missed inclusion

Intervention period Routine qSOFA calculation at ED triage

501 patients admitted with a suspected bacterial infection and screened for eligibility

24 patients meeting non-inclusion criteria Age <18 years: 10 Moribund at triage: 4

Pregnancy or breast-feeding: 9 In-ambulance antibiotic infusion: 1

390 included patients

40 excluded patients

No bacterial infection: 35 Patient refusal after emergency inclusion procedure: 3

Re-inclusion (2nd ED admission): 2

185 patients with sepsis and assessed for study endpoints

350 patients with definite bacterial infection

87 patients with missed inclusion

Fig. 1. Study flow-chart.

Table 1

Characteristics of the study population.

Characteristics	Baseline period - No qSOFA at ED triage	Intervention period - Routine qSOFA at ED triage	P value
	(n = 328)	(n = 350)
Sex, male	172 (52.4)	183 (52.3)	1.00
Age, years	69 (46-84)	69 (53-83)	0.73
Institutionalized patients	43 (13.1)	48 (13.7)	0.82
Hospital stay N 2 days within 90 days	47 (14.3)	43 (12.3)	0.49
Chronic diseases
Diabetes mellitus	66 (20.1)	75 (21.4)	0.71
Cardiac	133 (40.5)	117 (33.4)	0.06
Respiratory	46 (14.0)	58 (16.6)	0.39
Renal	25 (7.6)	43 (12.3)	0.05
Hepatic	8 (2.4)	6 (1.7)	0.59
Neurological	51 (15.5)	60 (17.1)	0.60
Psychiatric	20 (6.1)	33 (9.4)	0.12
Immunosuppressiona	67 (20.4)	70 (20.0)	0.92
MacCabe score 0	244 (74.4)	302 (86.3)	0.0002
1	72 (22.0)	37 (10.6)
2	12 (3.6)	11 (3.1)
Knauss score A-B	260 (79.3)	283 (80.8)	0.63
C-D	68 (20.7)	67 (19.2)
Antibiotic course prior to ED admissiona	49 (14.9)	56 (16.0)	0.75
Clinical parameters at triage
Systolic blood pressure, mmHg	132 (116-148)	130 (116-149)	0.92
Systolic blood pressure <= 100 mmHg	30 (9.1)	36 (10.3)	0.70
Diastolic blood pressure, mmHg	76 (66-87)	74 (64-86)	0.50
Mean blood pressure, mmHg	92 (82-104)	91 (81-103)	0.59
Heart rate, beats per min	96 (81-113)	98 (85-115)	0.06
Respiratory rate, breaths per min	20 (19-26)	22 (18-28)	0.43
Respiratory rate >= 22 breaths per min	136 (41.5)	174 (49.7)	0.04
Pulse oximetry - SpO2, %	96 (94-98)	96 (94-98)	0.63
Patients receiving oxygen	57 (17.4)	81 (23.1)	0.07
oxygen flow, litres per min	4 (3-9)	4 (3-9)	0.23
Glasgow coma scale	15 (15-15)	15 (15-15)	0.004
Glasgow coma scale <=13	12 (3.6)	21 (6.0)	0.21
Temperature, ?C	38.4 (38.0-39.0)	38.5 (38.0-39.1)	0.65
qSOFA score value at ED triage 0	172 (52.4)	163 (46.6)	0.09
1	135 (41.1)	146 (41.7)
2	20 (6.1)	38 (10.9)
3	1 (0.4)	3 (0.8)
>=2	21 (6.5)	41 (11.7)	0.02
Type of bacterial infection
Community-acquired	226 (68.9)	248 (70.9)	0.61
Healthcare-associated	102 (31.1)	102 (29.1)
Site of bacterial infection
Lower respiratory tract infection	108 (32.9)	145 (41.4)	0.03
Urinary tract infection	109 (33.2)	99 (30.2)	0.18
Abdominal infection	42 (12.8)	39 (11.9)	0.55
Skin and Soft tissues infection	48 (14.6)	31 (9.4)	0.02
Oropharyngeal suppuration	15 (4.6)	15 (4.6)	0.85
Othersa	27 (8.2)	32 (9.1)	0.68
More than one site of infection	19 (5.8)	11 (3.1)	0.13
positive blood culturesb	48 (14.6)	58 (16.6)	0.53
Microbiologically documented infection	132 (40.2)	143 (40.8)	0.88
arterial blood lactate level in the ED Value, mmol per litrec	1.3 (1.0-2.0)	1.2 (0.9-1.7)	0.25
Value N2 mmol per litre	29 (8.8)	30 (8.6)	1.00
Fluid resuscitation in the ED
Patients receiving at least one fluid challenge	33 (10.1)	36 (10.3)	1.00
Time from triage to the first fluid challenge, min	134 (60-360)	38 (20-69)	0.0003
Total administered volume, mL	1250 (500-2000)	1625 (1000-2000)	0.31
Vasopressor infusion in the ED	7 (2.1)	9 (2.6)	0.80
Antibiotic administration in the ED At least one administered dosea	241 (73.5)	267 (76.3)	0.42
Time from triage to the first dose, min	226 (127-343)	244 (130-356)	0.44
Number of administrated doses	1 (0-2)	1 (1-2)	0.34
SOFA score value at Day 1a	1 (0-2)	2 (0-3)	0.03
Criteria for sepsis at Day 1	151 (46.0)	185 (52.8)	0.08
Criteria for septic shock at Day 1	7 (2.1)	3 (0.8)	0.21
Patient status at ED discharge
Back to home	59 (18.0)	62 (17.7)	0.62
Hospital admission	269 (82.0)	287 (82.0)
Death in the ED SOFA score value at Day 2a	0 1 (0-2)	1 (0.3) 2 (0-3)	0.07
ICU admission

Table 1 (continued)
Characteristics	Baseline period - No qSOFA at ED triage (n = 328)	Intervention period - Routine qSOFA at ED triage (n = 350)	P value
<=48 h after triage	21 (6.4)	30 (8.6)	0.31
N48 h after triage	1 (0.4)	5 (1.4)	0.22
Length of hospital stay, days	7 (4-13)	6 (4-11)	0.65
In-hospital death at Day 7	11 (3.4)	12 (3.4)	1.00
In-hospital death, overall	24 (7.3)	19 (5.4)	0.35

Data are expressed as number of patients (%) or median values (1st-3rd interquartile range).

qSOFA, quick sepsis-related organ failure assessment; ED, emergency department; SOFA, sepsis-related organ failure assessment; ICU, intensive care unit.

^a Details are provided in Table S1 in the Appendix.

^b Sampled in 291 (88.7%) and 324 (92.6%) patients in the baseline and intervention periods, respectively.

^c Sampled in 134 (40.9%) and 178 (50.9%) patients in the baseline and intervention periods, respectively.

of patients with sepsis (Fig. S1). Among the 336 patients with sepsis over the two periods, 278 (82.7%) had a qSOFA value b2 at triage (Table S3). The sensibility and specificity of qSOFA for sepsis prediction were 17.3% (95% CI, 13.6%-21.7%) and 98.8% (95% CI, 97.0%-99.5%), re-

spectively, for an area under the Receiver operating characteristic curve of 0.72 (95% CI, 0.69-0.75) (Fig. 2 and Table S4). Patients with sepsis had higher ICU admission rate (49 [14.6%] versus 2 [0.6%] pa-

tients, P b 0.0001) and in-hospital mortality (33 [9.8%] versus 10 [2.9%] patients, P = 0.0002) than those without sepsis. Antibiotic use, fluid resuscitation and Vasopressor requirement according to the pres- ence of sepsis are exposed in Table S3 and Fig. S2.

Study endpoints in patients with sepsis

The proportion of patients who received a first antibiotic dose within one hour following triage was 7.3% (11/151) in the baseline period and 13.5% (25/185) in the intervention period (primary endpoint, absolute difference [AD] 6.2%, 95% CI [-0.5%, 12.7%], P = 0.08). The median time from triage to the first administered antibiotic dose (206 [102-336] versus 227 [115-345] min, P = 0.46) and the cumulative proportion of patients receiving a first antibiotic dose in the ED (Fig. 3, P = 0.95 by the log rank test) were similar in the two periods. No differ- ence was observed for any of the prespecified secondary endpoints, es- pecially the proportions of patients receiving a first antibiotic dose within three hours following triage (39.7% [50/151] versus 36.8% [68/ 185], AD -2.9%, 95% CI [-13.3%, 7.3%], P = 0.65), requiring ICU admis-

sion within 48 h following triage (13.2% [20/151] versus 15.7% [29/185], AD 2.4%, 95% CI [-5.5%, 2.9%], P = 0.64), or dying in the hospital (9.9% [15/151] versus 9.7% [18/185], AD -0.2%, 95% CI [-7.0%, 6.2%], P =

1.00) (Table 2).

Clinical significance of a qSOFA value >= 2

Patients with a qSOFA >= 2 at triage required ICU admission more commonly and had a higher in-hospital mortality rate than those with a qSOFA b 2 (Table 3). Sensibility, specificity, negative predictive value and positive predictive value of a qSOFA >= 2 to predict in-hospital mor- tality were 32.6% (95% CI, 20.5%-47.5%), 92.4% (95% CI, 90.1%-94.3%),

95.3% (95% CI, 93.3%-96.7%) and 22.6% (95% CI, 14.0%-34.4%), respec-

tively, with an AUROC of 0.68 (95% CI, 0.65-0.72) (Fig. 2). Patients with a qSOFA >= 2 received a first dose of antibiotic earlier than those with a qSOFA b 2 (154 [68-268] versus 241 [136-358] min, P = 0.0002); interestingly, this feature was observed in both the control and intervention periods (Table 4 and Fig. S3).

Discussion

In this study conducted in an ED with high visit volumes, the imple- mentation of a qSOFA-based triage procedure for admissions related to a bacterial infection was associated with a non-significant increase in the proportion of patients with sepsis receiving a first antibiotic dose within one hour following triage. No impact was observed on overall antibiotic timing, SOFA variations between triage and Day 2, or essential outcome measures such as the subsequent requirement for ICU admis- sion and in-hospital mortality.

qSOFA is an easily and repeatedly measurable Clinical score that has been introduced in 2016 for bedside screening and early identification of infected patients at-risk for prolonged ICU stay or in-hospital death [19]. External validation studies and meta-analyses published since the release of the Sepsis-3 conference statements yielded

Fig. 2. Receiver operating characteristic curves of qSOFA score values at triage for the prediction of sepsis at Day 1 (panel A) and in-hospital mortality (panel B) qSOFA, quick sepsis-related organ failure assessment; AUROC, area under the receiver operating characteristic curve; CI, confidence interval.

Fig. 3. Cumulative proportion of patients with sepsis receiving a first antibiotic dose in the emergency department in the baseline (no qSOFA calculation at triage) and intervention (routine qSOFA calculation at triage) periods qSOFA, quick sepsis-related organ failure assessment; ED, emergency department Kaplan-Meier curves are censored at the time of ED discharge of the last patient with sepsis on both periods.

heterogeneous data regarding the accuracy of a qSOFA value >= 2 for the prediction of sepsis-related fatality, with sensitivity and specificity ranging from 23% to 90% and 27% to 91%, respectively [20-24]. Yet, in a prospective cohort study conducted in 30 European EDs, the perfor- mances of qSOFA to portend in-hospital death were better than those of entities based on laboratory tests such as the systemic inflammatory response syndrome of the former Sepsis-2 definition of severe sepsis [25]. Besides this prognostic application, the score has been developed to prompt clinicians to initiate or escalate adequate sepsis-directed therapies - including antimicrobial agents - in patients with severe in- fection [8]. Triage alert protocols have been shown to lower time to an- tibiotic infusion in case of sepsis or septic shock [27,28]. However, to the best of our knowledge, the pragmatic input of qSOFA for sepsis manage- ment had not been appraised so far, especially in the ED. We hypothe- sized that a qSOFA-based triage protocol could have hastened the identification of patients admitted with a bacterial infection and subse- quently meeting the criteria for sepsis (i.e., before the availability of Lab- oratory results), thereby shortening the time to antibiotic initiation in this pivotal subgroup.

Patients with a qSOFA >= 2 received antibiotics more frequently and more precociously than those with a score value equal to 0 or 1.

Table 3

Correlation between qSOFA score values at triage and clinical outcomes.

	qSOFA value b 2	qSOFA value >= 2	P value
	(n = 616)	(n = 62)
SOFA score at Day 1	1 (0-3)	3 (2-6)	b0.0001
Criteria for sepsis at Day 1	278 (45.1)	58 (93.5)	b0.0001
Criteria for septic shock at Day 1	5 (0.8)	5 (8.1)	0.001
Hospital admission after ED evaluation	499 (81.0)	57 (91.9)	0.04
ICU admission within 48 h following triage	40 (6.5)	11 (17.7)	0.004
In-hospital death at Day 7	10 (1.6)	13 (21.0)	b0.0001
In-hospital death, overall	29 (9.2)	14 (22.6)	b0.0001

Data (both periods pooled) are expressed as number of patients (%) or median values (1st- 3rd interquartile range).

qSOFA, quick sepsis-related organ failure assessment; SOFA, sepsis-related organ failure assessment; ED, emergency department; ICU, intensive care unit.

Nevertheless, this feature was equally observed in the intervention and baseline periods, even though qSOFA was not part of triage proce- dures in the latter one. This suggests that warning clinical signs such as low blood pressure, elevated respiratory rate or altered mentation suffice to trigger the initiation of empirical antibiotic therapy regardless of the protocolized calculation of an established score. Owing to sea- sonal incidence variations, lower respiratory tract infections were more prevalent during the intervention period, which translated into a higher proportion of patients presenting with tachypnea and, conse- quently, a qSOFA >= 2 at triage. This result likely explains the trend to- wards a higher proportion of patients receiving a first antibiotic dose within one hour in the intervention period; still, this trend faded upon the second hour, and overall antibiotic timing was not affected by qSOFA-based triage.

In keeping with what has been reported in other populations of ED patients [29-31], we observed that the qSOFA value at triage had an ex- cellent specificity but a very poor sensitivity for early identification of patients who subsequently met the criteria for sepsis. This is a key ex- planation for the lack of significant impact of the study intervention on antibiotic timing and outcome. Importantly, clinical items at triage may not correspond to the worst values during the ED stay, which could have lessened the predictive accuracy of the score. Also, hepatic, hematological and renal dysfunctions were common in patients with sepsis: that these organ failures were not screened by the qSOFA score further reduced its sensitivity. Accelerating the management of ED pa- tients with sepsis implies the use of a highly sensitive rather than spe- cific screening tool at triage, and the performances of qSOFA may

Table 2

Pre-specified study endpoints in patients with sepsis.

Baseline period - No qSOFA at ED triage, n (%)

(n = 151)

Intervention period - Routine qSOFA at ED triage, n (%)

(n = 185)

Absolute difference,

%^a (95% CI)

P

value

Primary endpoint
First antibiotic dose within one hour following triage	11/151 (7.3)	25/185 (13.5)	6.2 (-0.5, 12.7)	0.08
Secondary endpoints
First adequate antibiotic dose within one hour following triage (patients with	3/61 (4.9)	8/82 (9.8)	4.9 (-4.9, 13.8)	0.35
microbiologically documented infection)
First antibiotic dose within three hours following triage	60/151 (39.7)	68/185 (36.8)	-2.9 (-13.3, 7.3)	0.65
First adequate antibiotic dose within three hours following triage (patients	21/61 (34.4)	19/82 (23.2)	-11.3 (-26.0, 3.5)	0.19
with microbiologically documented infection) Decrease in SOFA score >= 1 point between Day 1 and Day 2b	64/138 (46.4)	74/163 (45.4)	-1.0 (-12.1, 10.2)	0.91
ICU admission within 48 h following triage	20/151 (13.2)	29/185 (15.7)	2.4 (-5.3, 9.9)	0.64
Hospital LOS, days	11.6 +- 16.6	10.7 +- 13.0	-0.9 (-4.1, 2.3)	0.61
In-hospital death at Day 7	8/151 (5.3)	11/185 (5.9)	0.6 (-4.8, 5.7)	1.00
In-hospital death, overall	15/151 (9.9)	18/185 (9.7)	-0.2 (-7.0, 6.2)	1.00

Data are expressed as number of patients (%) or mean values +- standard deviation.

qSOFA, quick sepsis-related organ failure assessment; ED, emergency department; CI, confidence interval; SOFA, sepsis-related organ failure assessment; LOS, length of stay.

^a Except for hospital LOS (absolute difference and 95% CI expressed in days).

^b In patients still hospitalized at Day 2.

Table 4

Impact of qSOFA score values at triage on the timing of antibiotic initiation in the emergency department.

	qSOFA value b 2			qSOFA value >= 2
	Baseline period - No	Intervention period - Routine	P	Baseline period - No	Intervention period - Routine	P
	qSOFA at ED triage	qSOFA at ED triage	value	qSOFA at ED triage	qSOFA at ED triage	value
		(n = 309)			(n = 41)
	(n = 307)			(n = 21)
At least one antibiotic dose in the ED	222 (72.3)	231 (74.8)	0.52	19 (90.5)	36 (87.8)	1.00
Time from triage to the first	231 (131-350)	249 (140-366)	0.32	111 (81-245)	162 (61-280)	0.68
antibiotic dose, min
First antibiotic dose within one hour	16 (5.2)	22 (7.1)	0.40	2 (9.5)	9 (22.0)	0.30
following triage
First antibiotic dose within three	84 (27.4)	75 (24.3)	0.41	13 (61.9)	20 (48.8)	0.42
hours following triage

Data are expressed as number of patients (%) or median values (1st-3rd interquartile range). qSOFA, quick sepsis-related organ failure assessment; ED, emergency department.

compromise its relevance in this context. It remains to be investigated whether this new score may hasten the implementation of sepsis- directed cares in non-ED patients (e.g., for pre-hospital evaluation or in hospital wards).

This work has certain limitations that should be underlined. First, in both periods, the observed proportions of patients with sepsis receiving a first antibiotic dose within one hour after triage were lower than ex- pected for sample size calculation, as a probable consequence of sustainED patient flow and job strain for the ED staff. Indeed, an associ- ation between occupation rate and the delayed initiation of antibiotic therapy in patients with sepsis has been reported in other high- volume EDs [16,17]. This resulted in a substantial loss of statistical power. Hence, the benefit of a qSOFA-based triage procedure on antibi- otic timing cannot be firmly excluded, although this hypothesis appears unlikely given the poor sensitivity of qSOFA for early sepsis detection. Next, the single-center design of the study may limit its external valid- ity; however, patient characteristics, infection sites, the prevalence of sepsis and mortality rates were concordant with those reported in other ED cohorts [11,25]. Third, respiratory rate at triage was not avail- able for a restricted subset of patients from both periods; however, mul- tiple imputations showed that these missing values for the respiratory item of qSOFA exerted no effect on overall results (Electronic supple- mentary material). Fourth, the time from triage to the first clinical eval- uation by an ED physician was not measured. Lastly, follow-up ended at hospital discharge; still, the lack of difference for study endpoints makes unlikely a benefit of qSOFA-based triage on later outcomes (e.g., at Day 28).

Conclusion

The implementation of a qSOFA-based triage procedure did not im- prove antibiotic timing and outcomes in patients with sepsis admitted to an ED with high visit volumes. However, the study may have been underpowered. The low sensitivity of initial qSOFA values for the pre- diction of sepsis does not support the use of this score for triage of ED patients presenting with clinical signs suggesting bacterial infection.

Author contributions

J.P., J.P., T.B. and F.B. designed the study, analyzed the data and wrote the article; all other authors contributed to patient enrolment and data collection and revised the draft critically for important intellectual content.

Ethics approval and consent to participate

The study protocol was approved by the CPP Marseille/Sud- Mediterranee I ethic committee on June 8, 2017 (IRB no. 2017-

A01000-53). Procedures to obtain patient consent are detailed in the Appendix.

Declaration of competing interests

All authors declare that they have no conflict of interest related to the present study.

Availability of data and material

The datasets used and analyzed during the current study are avail- able from the corresponding author on request.

Funding

This work was funded by the Centre Hospitalier Regional d’Orleans, Orleans, France.

Acknowledgements

The authors thank the ED nurses for their essential contribution to the study.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi. org/10.1016/j.ajem.2019.05.022.

References

1. Cecconi M, Evans L, Levy M, Rhodes A. Sepsis and septic shock. Lancet 2018;392 (10141):75-87.
2. Fleischmann C, Scherag A, Adhikari NK, Hartog CS, Tsaganos T, Schlattmann P, et al. Assessment of global incidence and mortality of hospital-treated sepsis: current es- timates and limitations. Am J Respir Crit Care Med 2016;193(3):259-72.
3. World Health Organization. 7th world health assembly. Improving the prevention, diagnosis and clinical management of sepsis. Report A70/13, April. available at www.who.int; 2017.
4. Yende S, Austin S, Rhodes A, Finfer S, Opal S, Thompson T, et al. Long-term quality of life among survivors of severe sepsis: analyses of two international trials. Crit Care Med 2016;44(8):1461-7.
5. Iwashyna TJ, Ely EW, Smith DM, Langa KM. Long-term cognitive impairment and functional disability among survivors of severe sepsis. JAMA 2010;304(16): 1787-94.
6. Ou SM, Chu H, Chao PW, Lee YJ, Kuo SC, Chen TJ, et al. Long-term mortality and major adverse cardiovascular events in sepsis survivors: a nationwide population- based study. Am J Respir Crit Care Med 2016;194(2):209-17.
7. Tiru B, DiNino EK, Orenstein A, Mailloux PT, Pesaturo A, Gupta A, et al. The economic and humanistic burden of severe sepsis. Pharmacoeconomics 2015;33(9):925-37.
8. Singer M, Deutschman CS, Seymour CW, Shankar-Hari M, Annane D, Bauer M, et al. The third international consensus definitions for sepsis and septic shock (Sepsis-3). JAMA 2016;315(8):801-10.
9. 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(9):1907-15.
10. de Groot B, Ansems A, Gerling DH, Rijpsma D, van Amstel P, Linzel D, et al. The asso- ciation between time to antibiotics and relevant clinical outcomes in emergency de- partment patients with various stages of sepsis: a prospective multi-center study. Crit Care 2015;19:194.
11. 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(7):856-63.
12. 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. N Engl J Med 2017;376(23):2235-44.
13. Ferrer R, Martin-Loeches I, Phillips G, Osborn TM, Townsend S, Dellinger RP, et al. Empiric antibiotic treatment reduces mortality in severe sepsis and septic shock from the first hour: results from a guideline-based performance improvement pro- gram. Crit Care Med 2014;42(8):1749-55.
14. Zhang D, Micek ST, Kollef MH. Time to appropriate antibiotic therapy is an indepen- dent determinant of post-infection ICU and hospital lengths of stay in patients with sepsis. Crit Care Med 2015;43(10):2133-40.
15. Rhodes A, Evans LE, Alhazzani W, Levy MM, Antonelli M, Ferrer R, et al. Surviving sepsis campaign: international guidelines for management of sepsis and septic shock, 2016. Intensive Care Med 2017;43(3):304-77.
16. Peltan ID, Bledsoe JR, Oniki TA, Sorensen J, Jephson AR, Allen TL, et al. Emergency de- partment crowding is associated with delayed antibiotics for sepsis. Ann Emerg Med 2018. https://doi.org/10.1016/j.annemergmed.2018.10.007.
17. Gaieski DF, Agarwal AK, Mikkelsen ME, Drumheller B, Cham Sante S, Shofer FS, et al. The impact of ED crowding on early interventions and mortality in patients with se- vere sepsis. Am J Emerg Med 2017;35(7):953-60.
18. Wallgren UM, Antonsson VE, Castren MK, Kurland L. Longer time to antibiotics and higher mortality among septic patients with non-specific presentations-a cross sec- tional study of emergency department patients indicating that a screening tool may improve identification. Scand J Trauma Resusc Emerg Med 2016;24(1).
19. Seymour CW, Liu VX, Iwashyna TJ, Brunkhorst FM, Rea TD, Scherag A, et al. Assess- ment of clinical criteria for sepsis: for the third international consensus definitions for sepsis and septic shock (Sepsis-3). JAMA 2016;315(8):762-74.
20. Maitra S, Som A, Bhattacharjee S. Accuracy of quick Sequential Organ Failure Assess- ment (qSOFA) score and Systemic Inflammatory Response Syndrome criteria for predicting mortality in hospitalized patients with suspected infection: a meta- analysis of observational studies. Clin Microbiol Infect 2018;24(11):1123-9.
21. Fernando SM, Tran A, Taljaard M, Cheng W, Rochwerg B, Seely AJE, et al. Prognostic accuracy of the quick sequential organ failure assessment for mortality in patients

    with suspected infection: a systematic review and meta-analysis. Ann Intern Med 2018;168(4):266-75.

    Tan TL, Tang YJ, Ching LJ, Abdullah N, Neoh HM. Comparison of prognostic accuracy of the quick sepsis-related organ failure assessment between short- & long-term mortality in patients presenting outside of the intensive care unit: a systematic re- view & meta-analysis. Sci Rep 2018;8(1):16698.

22. Song JU, Sin CK, Park HK, Shim SR, Lee J. Performance of the quick sequential (sepsis- related) organ failure assessment score as a Prognostic tool in infected patients out- side the intensive care unit: a systematic review and meta-analysis. Crit Care 2018; 22(1):28.
23. Serafim R, Gomes JA, Salluh J, Povoa P. A comparison of the quick-SOFA and systemic inflammatory response syndrome criteria for the diagnosis of sepsis and prediction of mortality: a systematic review and meta-analysis. Chest 2018;153(3):646-55.
24. Freund Y, Lemachatti N, Krastinova E, Van Laer M, Claessens YE, Avondo A, et al. Prognostic accuracy of Sepsis-3 criteria for in-hospital mortality among patients with suspected infection presenting to the emergency department. JAMA 2017; 317(3):301-8.
25. Churpek MM, Snyder A, Han X, Sokol S, Pettit N, Howell MD, et al. Quick sepsis- related organ failure assessment, systemic inflammatory response syndrome, and early warning scores for detecting clinical deterioration in infected patients outside the intensive care unit. Am J Respir Crit Care Med 2017;195(7):906-11.
26. Hayden GE, Tuuri RE, Scott R, Losek JD, Blackshaw AM, Schoenling AJ, et al. Triage sepsis alert and sepsis protocol lower times to fluids and antibiotics in the ED. Am J Emerg Med 2016;34(1):1-9.
27. Gatewood MO, Wemple M, Greco S, Kritek PA, Durvasula R. A quality improvement project to improve early sepsis care in the emergency department. BMJ Qual Saf 2015;24(12):787-95.
28. Williams JM, Greenslade JH, McKenzie JV, Chu K, Brown AFT, Lipman J. Systemic in- flammatory response syndrome, qSOFA, and organ dysfunction: insights from a pro- spective database of ED patients with infection. Chest 2017;151(3):586-96.
29. Giamarellos-Bourboulis EJ, Tsaganos T, Tsangaris I, Lada M, Routsi C, Sinapidis D, et al. Validation of the new Sepsis-3 definitions: proposal for improvement in early risk identification. Clin Microbiol Infect 2017;23(2):104-9.
30. Askim A, Moser F, Gustad LT, Stene H, Gundersen M, Asvold BO, et al. Poor perfor- mance of qSOFA score in predicting severe sepsis and mortality - a prospective study of patients admitted with infection to the emergency department. Scand J Trauma Resusc Emerg Med 2017;25(1):56.