Impact of Prehospital Mobile Intensive Care Unit Intervention on Mortality of Patients with Sepsis

Abstract

Objective

The outcome of sepsis relies on the early diagnosis and implementation of appropriate treatments. For management of out-of-hospital patients with sepsis, prehospital emergency services, named Service d’Aide Médicale d’Urgence (SAMU) in France, dispatch to the scene an emergency mobile team (EMT) or a mobile intensive care unit (MICU) based on the patient’s severity. Therefore, patients are admitted to the emergency department (ED) or to the intensive care unit (ICU). The impact of MICU intervention on patient’s prognosis remains unclear. The aim of the present study was to describe the impact of MICU intervention on mortality on day 28 (D28) of patients with sepsis.

Methods

We performed a retrospective study on patients with sepsis managed by prehospital teams, MICU or EMT, before admission to the ED or ICU. The primary outcome was mortality on D28.

Results

The SAMU received 30,642 calls during the study period with 140 patients with suspected sepsis. The suspected origin of sepsis was mainly pulmonary for 78 (55%) patients. Thirteen (9%) patients died on D28, 12 in the ED and 1 in the ICU. Two patients were admitted to the hospital by a MICU. After adjusting for confounding factors, the relative risk of mortality on D28 for patients admitted to the hospital by a MICU was 0.40.

Conclusion

We describe an association between MICU intervention and mortality on D28. MICU intervention for out-of-hospital patients with sepsis is associated with 60% reduced mortality on D28. Larger studies are needed to confirm the impact of the intervention of MICU on mortality of patients with sepsis.

Introduction

Septic shock concerns approximately 19 million people worldwide annually (1, 2), generating 180,000 deaths in the United States (3). The in-hospital mortality rate related to septic shock still reaches 30% on day 28 (D28) (2, 4) despite continuous guidelines to improve the management of these patients within the last two decades (5). It is now clearly admitted that outcome relies on the early identification of patients with sepsis at high risk of poor outcome to initiate rapid implementation of appropriate treatments, especially haemodynamic optimisation and antibiotic administration (6–8). For out-of-hospital patients, sepsis has to be diagnosed as early as possible starting in the prehospital setting.

In France, the management of out-of-hospital emergencies is based on the emergency medical services called Service d’Aide Médicale d’Urgence (SAMU). The SAMU is reached via a national access call number, the number 15 (9). For patients requiring hospital admission, an emergency mobile team (EMT), a regular ambulance or a prehospital mobile intensive care unit (MICU) is dispatched to the scene. While an EMT always transfers patients to the emergency department (ED), patients transported by a MICU might be admitted to the ED or to the intensive care unit (ICU) according to their severity. Thus, efficient prehospital triage is crucial to enable appropriate orientation of patients. However, relevant tools to diagnose sepsis and to evaluate patients’ severity are lacking in this context. In fact, studies reported the lack of efficiency of the quick sepsis-related organ dysfunction assessment score (qSOFA) in the prehospital setting to predict ICU admission (10–14). Therefore, in this context, the evaluation of patients remains complex. Further efforts are needed to improve the early triage of patients with sepsis in this environment. Positive impact of early specialised management of patients by prehospital MICU intervention was shown for cardiac arrest and severe trauma (9). In the case of sepsis, the purpose of early management of patients by a MICU has not been evaluated yet.

The aim of the present study was to describe the impact of prehospital MICU intervention on mortality on D28 of patients with sepsis.

Methods

Population and data

Patients with suspected sepsis cared for by the SAMU of Paris (SAMU 75) between 1 April 2011 and 31 May 2011 were included in the study. Patients <18 years, pregnant women and patients with incomplete data sets were excluded from the study.

Sepsis was suspected in case of fever ≥38.0°C during 1 h or 38.3°C once or hypothermia <36°C with a medical history compatible with an infection (15, 16). The site of infection was suspected according to the patient’s medical history and clinical signs.

Data including patient’s demographic characteristics and clinical evaluation (systolic, diastolic and mean blood pressure (SBP, DBP and MBP), heart rate (HR), respiratory rate, temperature, Glasgow Coma Scale (GCS) and pulse oximetry) were retrieved from recorded phone calls made during prehospital care delivery and from the SAMU prehospital medical files. The primary outcome was mortality on D28 and was collected from hospital records.

To minimise the bias associated with data abstraction, data collection was performed by a single investigator using a standardised abstraction template, defined prior to data collection (17).

Immunocompromised status was retained if one or more of the following items were present in the patient’s medical history: diabetes mellitus, chronic renal insufficiency, corticosteroids or another immunosuppressive treatment and infection by human immunodeficiency virus and/or viral hepatitis C.

Assessment of the prehospital qSOFA score was based on the three following clinical features: SBP ≤100 mmHg, RR ≥22 breaths min⁻¹ and altered mental status determined by a GCS <15, with one point awarded for each item.

The local ethics committee of Comité de Protection des Personnes, Ile de France 2, Paris- France approved the study in accordance with the French legislation (Number: 2015-08-03-SC). Consent from patients was waived for participation in this observational study.

Study design and setting

The SAMU is composed of switchboard operators and physicians who determine, over the phone, the appropriate level of care to dispatch to the scene. The decision-making is based on the patient’s medical history and symptoms as related by the patient himself, his relatives or any witness.

For life-threatening emergencies, the Service Mobile d’Urgence et de Réanimation, corresponding to a prehospital MICU, enables direct admission to the ICU or to the ED (18). The MICU is composed of a driver, a nurse and an emergency physician and is equipped with medical devices and drugs allowing the initial management of main organ deficiencies (9). In the case of sepsis, the MICU is able to initiate haemodynamic optimisation (fluid expansion and/or catecholamine infusion) and antibiotherapy. For less severe cases, an EMT (firefighters) or an ambulance, corresponding to a paramedic team, is dispatched to the scene.

After arrival to the scene of the appropriate care support, EMT or MICU, the patient’s demographic characteristics and clinical evaluation (SBP, DBP and MBP; HR; rr; body temperature; GCS and pulse oximetry) are communicated to the regulation call centre to decide on the best course of action. Thereafter, patients are transferred either to the ED or to the ICU.

Statistical analysis

Univariate and multivariable analyses were conducted to evaluate the relationship between all covariates and mortality on D28.

A propensity score analysis, including age >80 years, immunocompromised status and qSOFA ≥2, was used to reduce the effect of confounders (19). Age >80 years, immunocompromised status and qSOFA ≥2 were reported to be associated with increased mortality in patients with sepsis (20–22).

The covariate balancing propensity score method was used to allow simultaneous optimisation of the prediction of the outcome and the covariate balance. As p-value is influenced by the sample size, standardised mean difference was used to evaluate imbalance matching using the following formulae described by Austin (23):

$$ASD=100*\frac{\text{x\hspace{0.17em}}(\text{TT})\text{-x\hspace{0.17em}}(\text{CTL})}{\sqrt{\frac{{\text{s}}^2(TT)+s^2(CTL)}{2}}},$$

where x corresponds to the mean or proportion for binary variables and classes of categorical variables, and s is the variance.

Cases correspond to patients admitted to the ED or to the ICU by a MICU, and controls correspond to patients admitted to the ED by an EMT. In the matched sample, baseline characteristics between cases and controls for mortality on D28 were compared using an unpaired t-test. All p-values were two-tailed. A p-value <0.05 was considered significant.

At least, the relative risk (RR) was used to evaluate patients’ mortality on D28 transported by a MICU or an EMT.

Data are expressed as mean±standard variation for Gaussian variables, median with 1st quartile and 3rd quartile for non-normal variables or number and percentage (%).

All analyses were performed using R 3.4.2 (http://www.R-project.org; the R Foundation for Statistical Computing, Vienna, Austria).

Results

During the study period, 30,642 calls were received by the SAMU 75 call centre. Of the 30,642 calls, 140 concerned patients with suspected sepsis (Figure 1).

Figure 1.

Flowchart

SAMU: Service d’Aide Médicale d’Urgence; MICU: mobile intensive care unit; EMT: emergency mobile team

Of the 140 patients, 76 (54%) were male, and the median age was 72 (43–81) years (Table 1). Forty (29%) patients were >80 years. Eighty-six (61%) patients were immunocompromised.

Table 1.

Demographic and clinical characteristics of patients with sepsis admitted to the hospital by a MICU or an EMT

Variables	Alive on day 28 (n=127)	Deceased on day 28 (n=13)	Overall population (n=140)
Age (years)	70 (39–81)	78 (66–87)	72 (43–81)
Age >80 years	36 (90%)	4 (10%)	40 (29%)
Male gender	67 (52%)	9 (69%)	76 (54%)
Immunosuppression	77 (61%)	9 (69%)	86 (61%)
Systolic blood pressure (mmHg)	120 (96–138)	101 (95–110)	110 (96–130)
Diastolic blood pressure (mmHg)	70 (50–80)	60 (50–70)	70 (50–80)
Mean blood pressure (mmHg)	87 (63–100)	73 (65–83)	83 (63–100)
Heart rate (beats min−1)	108 (97–120)	101 (81–130)	107 (96–120)
Pulse oximetry (%)	96 (90–98)	93 (90–96)	96 (90–98)
Respiratory rate (breaths min−1)	24 (22–28)	26 (24–30)	25 (22–28)
GCS	15 (15–15)	15 (15–15)	15 (15–15)
Body temperature (°C)	38.7 (37.9–39.0)	38.5 (37.0–39.0)	38.6 (37.9–39.0)
Prehospital management by MICU	14 (11%)	2 (15%)	16 (11%)
Patients admitted in the ED	112 (95%)	6 (5%)	118 (84%)
Patients admitted in the ICU	15 (68%)	7 (32%)	22 (16%)
qSOFA >2	39 (31%)	7 (54%)	45 (32%)

Quantitative variables are expressed as median (1^st quartile–3^rd quartile). Qualitative variables are expressed as absolute value and percentage (%).

GCS: Glasgow Coma Scale; qSOFA: quick sepsis-related organ dysfunction assessment score; MICU: mobile intensive care unit; EMT: emergency mobile team; ED: emergency department; ICU: intensive care unit

The predominant sites of infection were pulmonary for 77 (55%) patients, urinary for 20 (14%) patients and abdominal for 19 (13%) patients.

In the prehospital setting, median temperature was 38.6°C (37.9–39.0), median SBP was 110 (96–130) mmHg, and median respiratory rate was 25 (22–28) breaths min⁻¹. Forty-five (32%) patients had a qSOFA ≥2. All patients had a GCS of 15.

Among the 140 patients with suspected sepsis in the prehospital setting, 16 (11%) were transferred to the hospital by a MICU. None of them received antibiotics prior to hospital admission. Thirteen (9%) patients died on D28, 12 had been admitted to the ED, 1 to the ICU, and 2 had been transported by a MICU. Among the 2 deceased patients transported by a MICU, 1 had been admitted to the ED, and 1 to the ICU.

All deaths were related to a documented history of sepsis. A microbiological documentation was obtained in 42% of the cases and was pulmonary for 38 (65%) patients (Table 2).

Table 2.

Presumed site of infection in patients with sepsis admitted to the hospital by a MICU or an EMT

Site of infection	n	%
Pulmonary	77	55
Urinary	20	14
Digestive	19	13
Undefined	10	8
Meningeal	6	4
Cutaneous	4	3
Ear–nose–throat	3	2
Bone	1	1

Data are expressed as absolute value with percentage (%). MICU: mobile intensive care unit; EMT: emergency mobile team

In the univariate analysis, no variable was significantly associated with death on D28 (Table 3). No variables included in the propensity score significantly differed between the cases and the controls after propensity score matching (Figure 2 and Table 4).

Table 3.

Univariate analysis of factors associated with mortality on day 28 in patients with sepsis admitted to the hospital by a MICU or an EMT

	Univariate analysis
Variables	OR (95% CI)	p
Age (years)	1.03 (0.99–1.06)	0.10
Age >80 years	1.12 (0.29–3.69)	0.85
Male gender	2.01 (0.62–7.75)	0.26
Immunosuppression	1.46 (0.45–5.63)	0.55
Systolic blood pressure (mmHg)	0.99 (0.95–1.01)	0.06
Diastolic blood pressure (mmHg)	0.98 (0.95–1.01)	0.24
Mean blood pressure (mmHg)	0.98 (0.94–1.01)	0.12
Heart rate (beats min−1)	1.01 (0.98–1.03)	0.59
Pulse oximetry (%)	0.99 (0.94–1.07)	0.79
Respiratory rate (breaths min−1)	1.06 (0.97–1.15)	0.19
GCS	1.01 (0.58–2.67)	0.98
Body temperature (°C)	0.75 (0.51–1.13)	0.14
MICU	1.47 (0.21–6.23)	0.64
qSOFA >2	1.93 (0.59–6.19)	0.26

Data are presented as p-value and OR with a 95% CI. GCS: Glasgow Coma Scale; MICU: mobile intensive care unit; EMT: emergency mobile team; qSOFA: quick sepsis-related organ dysfunction assessment score; OR: odds ratio; 95% CI: 95% confidence interval

Figure 2.

Standard mean deviation between cases and controls after propensity score matching. Cases correspond to patients admitted to the hospital by a MICU, and controls correspond to patients admitted by an EMT

qSOFA: quick sepsis-related organ dysfunction assessment score; MICU: mobile intensive care unit; EMT: emergency mobile team; IS: immunosuppression

Table 4.

Comparison of variables predicting mortality on day 28 of patients with sepsis included in the propensity score before and after matching

PS covariate	Before matching (n=140)		p (d*)	After matching (n=32)		p (d*)
	Cases n=16	Controls n=124		Cases n=16	Controls n=16
Age >80 years	6 (38%)	34 (27%)	0.21	6 (38%)	10 (63%)	<10–3
qSOFA >2	7 (44%)	38 (31%)	0.27	7 (44%)	7 (44%)	<10–3
IS	11 (69%)	75 (60%)	0.17	11 (69%)	11 (69%)	<10–3

Values are expressed as number and percentage (%). Cases correspond to patients admitted to the hospital by a MICU, and controls correspond to patients admitted by an EMT.

PS: propensity score; qSOFA: quick sepsis-related organ dysfunction assessment score; IS: immunosuppression; MICU: mobile intensive care unit; EMT: emergency mobile team

After adjusting for confounding factors, the RR of mortality on D28 regarding patients with presumed sepsis admitted to the hospital by a MICU reached 0.40, whereas it was 2.5 for those admitted to the hospital by an EMT.

Discussion

In the present study, we report an association between mortality on D28 and intervention of a MICU for patients with sepsis cared for in the prehospital setting. The intervention of a prehospital MICU is associated with 60% reduced mortality on D28 (RR=0.4). To our knowledge, this work is the first study to investigate the role of prehospital transportation in the prognosis of patients with sepsis. Interhospital transportation of critically ill patients by a MICU was widely described showing clinical stability based on the SOFA score (24). On the contrary, prehospital transportation raises other issues as the level of emergency has to be first identified by the dispatch centre. In fact, a significant proportion of patients are transported by an inappropriate mode of transport. Half of out-of-hospital patients with sepsis are transported by non-EMS (25). The relationship with mortality is controversial as some found no impact of prehospital transportation (25), whereas others described increased mortality of patients transported by EMS (26). On the other hand, patients with sepsis transported by EMS died more often as they were more seriously ill (26) and transported with less urgent rides than needed (27). However, appropriate EMS transportation shortened the time to initiation of antibiotics and fluid expansion of approximately 30–60 min (25, 28).

The time lapse between the diagnosis of sepsis and the initiation of treatments is a well-proven prognostic factor affecting patients’ outcome in sepsis (29–32). Therefore, improvement of the management of sepsis in the survival chain, in analogy to cardiac arrest management, aimed at reducing the time to treatment (30). In fact, early ICU admission significantly improved the prognosis of patients with septic shock (33). However, no defined strategy is currently used for management of prehospital patients with sepsis, contrary to the clear guidelines on acute coronary syndromes (34, 35). Should critical patients be admitted to the ICU directly or should they first transit through the ED? Should they be cared for in the prehospital setting directly by a MICU or might an EMT be sufficient? The absence of relevant tools for prehospital triage of patients with sepsis underlines the complexity of the issue. In fact, no vital sign is enough for decision-making. Additionally, biological values are time-consuming and hard to achieve in this context. These difficulties account for differences between studies on the benefit of the bundle of care strategy (29, 30, 32). Anyhow, the bundle of care concept describes key points for global management of patients with sepsis that must be performed within a time scale of 3 h to significantly decrease mortality (29).

Strengths and limitations

Some limitations restrict the interpretation of our results. First, this is a retrospective and single-centre study. Therefore, our results may not be extrapolated to other populations, medical institutions or countries. In fact, studies on prehospital transportation mode compared EMS with non-EMS without including MICUs (25, 27, 28). Moreover, the quality of care provided by EMS significantly differed between hospitals and countries. In many countries, paramedical teams have an insufficient knowledge regarding sepsis with an under evaluation of hypothermia (36) and poor recognition of sepsis (37). In parallel, in some countries, EMS teams have the ability to recognise sepsis and thus reduce the time to initiation of optimal therapies. Second, the sample size is small, decreasing the power of the study. Third, a sorting bias may exist as data were retrospectively collected from the SAMU regulation centre call recorded phone calls and ICU or ED medical reports. Finally, we could not investigate the effect of prehospital antibiotherapy as no patients received antibiotics prior to hospital admission. Similarly, we could not study haemodynamic optimisation as data lacked from the SAMU reports. Importantly, prehospital transportation in Paris is very efficient, and the time required to transfer a patient to the hospital is less likely to delay the time to treatment or interfere with mortality. Anyhow, antibiotic administration is a confounding factor and would have interfered with our results.

Despite these limitations, we observed a positive impact of the intervention of prehospital MICU on mortality of patients with sepsis on D28 initially cared for in the prehospital setting. Further larger cohorts are needed to confirm these preliminary data and to provide a causal link between mortality and prehospital MICU intervention.

Conclusion

We described an association between prehospital MICU intervention and mortality on D28 in patients with sepsis. MICU transportation of out-of-hospital patients with sepsis was associated with a RR reduction of 60% of death on D28.

These results underline the need to optimise prehospital management of patients with sepsis starting from patients’ transportation to hospital admission. Larger studies are needed to confirm the impact of MICU intervention on mortality of patients with sepsis initially cared for in the prehospital setting.