1. Introduction

Septic shock is the most severe form of sepsis and is described as persistent hypotension requiring vasopressors to maintain a mean arterial pressure (MAP) of 65 mmHg despite adequate volume resuscitation. The risk of death increases by 5.3% for every hour that vasopressor initiation is delayed, similar to the time-dependent risk of delayed antimicrobials in sepsis [1, 2]. The Surviving Sepsis Campaign recommends including vasopressor initiation in the crucial 1-h bundle for fluid-resistant hypotension, with norepinephrine recommended as a first-line vasopressor [3]. The in-hospital mortality rate of patients with septic shock is higher when vasopressors are delayed [4]. Vasopressor initiation is delayed beyond 4 hours with a four-fold increase in the odds of worsening organ failure (odds ratio (OR) 4.34, 95% confidence interval (CI) 1.47–12.79, p = 0.008) than those who receive vasopressors within 4 hours [5]. Hence, timely vasopressors initiation is crucial for effective septic shock management.

Septic shock requiring high norepinephrine levels result in impaired catecholamine responsiveness and uncontrolled vasoplegia due to receptor signaling changes, metabolic derangements and depletion of endogenous vasoactive hormones [6]. Excess norepinephrine stimulation may result in ischemic digits, splanchnic hypoxia, necrosis and serious morbidity. A mortality rate of 60% to 80% was shown in septic shock patients receiving 1 g/kg/min of norepinephrine equivalent [7, 8]. Based on a retrospective study of 324 septic shock patients, the average death rate was 48%, while mortality reached 90% for patients receiving more than 1 μg/kg/min of norepinephrine [9].

When escalated norepinephrine dosage failed to maintain MAP targets in some patients, other vasopressors were initiated. Extremely high doses of norepinephrine may induce a relative catecholamine-refractory state, which can be treated with additional vasopressors with different receptors to maintain adequate perfusion pressures and mitigate progressive multiorgan failure [2, 10]. In patients who require high vasopressor doses, it is physiologically rational to combine multiple vasopressors as part of multimodal therapy targeting multiple receptor [6]. However, the optimal norepinephrine dosage at which additional vasopressors are initiated remains unknown [11]. Septic shock patients given additional vasopressin at <15 μg/min norepinephrine during the Vasopressin and Septic Shock Trial (VASST) had lower 28-day and 90-day mortality [12]. Researchers found that for every 10 μg/min increase in the norepinephrine-equivalent dose at the time of vasopressin initiation in septic shock patients, the odds of in-hospital mortality increased by 20.7% [13].

Early initiation of a vasopressor is clearly better than later initiation, but the optimal timing of a secondary agent is less clear. Therefore, this study examined the association between timing of multimodal vasopressor therapy and clinical outcomes in septic shock patients receiving a maximum norepinephrine-equivalent dose of >0.2 μg/kg/min therapy. In high-dose vasopressor-dependent patients with septic shock, earlier initiation of multimodal vasopressor therapy was hypothesized to improve prognosis.

2. Methods

2.2 Study population and definitions

In 57,328 non-repetitive ICU admissions, sepsis was diagnosed according to the Angus criteria [15]. Septic shock patients receiving multiple vasopressors during ICU stay were included. Clinical vasopressors include norepinephrine, epinephrine, dopamine, phenylephrine and vasopressin. We calculated norepinephrine-equivalent vasopressor doses for epinephrine, dopamine, phenylephrine and vasopressin as previously described [16]. Since norepinephrine is recommended as first-line therapy in septic shock [3], patients given other vasopressors prior to norepinephrine or without norepinephrine usage were excluded. Septic shock patients receiving a maximum norepinephrine-equivalent dose of less than 0.2 μg/kg/min were also excluded. Finally, 966 patients included in this study were divided into three groups: (1) Early multimodal vasopressor therapy (EMMVT) group (additional vasopressors administered when norepinephrine dose ≤0.2 μg/kg/min); (2) Later multimodal vasopressor therapy (LMMVT) group (additional vasopressors administered when norepinephrine between 0.2–0.5 μg/kg/min); (3) Delayed multimodal vasopressor therapy (DMMVT) group (additional vasopressors administered when norepinephrine dose ≥0.5 μg/kg/min) (Fig. 1). A cutoff of 0.2 and 0.5 μg/kg/min was used since over 0.2 μg/kg/min of norepinephrine was previously defined as high-dose vasopressor [17]. Norepinephrine requirement more than 0.5 μg/kg/min was defined as refractory shock [18].

Fig. 1.Flow diagram describing the screening, recruitment of patients. ICU: intensive care unit.

The following data were extracted: patients’ baseline characteristics, including sex, age, admission type, comorbidity, support therapies on admission (including mechanical ventilation and renal replacement therapy), Sequential Organ Failure Assessment (SOFA) scores, time of ICU and hospital admission and discharge, and the date of death. A 28-day mortality was the primary endpoint. 90-day mortality, ICU and hospital mortality, as well as the length of ICU and hospital stay were secondary endpoints.

3. Results

3.2 Primary analysis

A survival curve analysis was performed to investigate the association between multimodal vasopressor timing and 28-day mortality. LMMVT (hazard ratio, 1.849; p < 0.001) and DMMVT (hazard ratio, 2.329; p < 0.001) were significantly correlated with 28-day mortality (Fig. 2C). Additionally, we included age, admission type, comorbidities, SOFA score and multimodal vasopressor therapy in a multivariable Cox proportional hazard model. Compared to EMMVT, LMMVT (hazard ratio, 1.643; p < 0.001) and DMMVT (hazard ratio, 2.192; p < 0.001) were associated with increased risk of 28-day mortality (Table 2). Several other covariates were significantly associated with 28-day mortality, including older age, emergency and urgent admission, SOFA score, metastatic cancer, and liver diseases. However, obesity was a protective factor (Table 2).

Fig. 2.Comparisons of clinical outcomes and vasopressor usage between survivors and non-survivors. (A) Norepinephrine dose at the second vasopressor initiation between 28-day survivors and non-survivors. Outliers were not presented in boxplot. (B) Numbers of vasopressors used among survivors and non-survivors. Vasopressors including norepinephrine, epinephrine, dopamine, phenylephrine and vasopressin were considered. (C) Kaplan-Meier curves showing the association between the multimodal vasopressor therapy timing and the 28-day mortality. EMMVT: early multimodal vasopressor therapy; LMMVT: later multimodal vasopressor therapy; DMMVT: delayed multimodal vasopressor therapy; ICU: intensive care unit.

Table 2.Cox proportional hazard models exploring the association between the time of multimodal vasopressor and 28-day mortality.

Factors	Hazard ratio	95% CI	p value
Age (yr, n, %)
	18–50	Reference	Reference	Reference
	50–60	1.679	1.231–2.289	0.001
	60–70	1.947	1.442–2.628	<0.001
	70–80	2.054	1.515–2.785	<0.001
	>80	2.966	2.206–3.987	<0.001
Admission type
	Elective	Reference	Reference	Reference
	Emergency	1.755	1.065–2.890	0.027
	Urgent	2.308	1.138–4.682	0.020
Metastatic cancer	2.365	1.754–3.191	<0.001
Liver diseases	1.405	1.096–1.800	0.007
Obesity	0.620	0.410–0.939	0.024
SOFA score	1.106	1.081–1.133	<0.001
Multimodal vasopressor therapy
	EMMVT	Reference	Reference	Reference
	LMMVT	1.643	1.331–2.028	<0.001
	DMMVT	2.192	1.684–2.852	<0.001
SOFA: sequential organ failure assessment; CI: Confidence interval; EMMVT: early multimodal vasopressor therapy; LMMVT: later multimodal vasopressor therapy; DMMVT: delayed multimodal vasopressor therapy.

A second model included the interaction between multimodal vasopressor groups and SOFA categories, since we hypothesized that the effect of multimodal vasopressor time on survival might differ depending on illness severity. Patients were stratified according to SOFA score quartiles. For patients with SOFA score <8 (p < 0.001), SOFA score 10–13 (0.006) and SOFA score >13 (0.030), later or delayed multimodal vasopressor therapy was associated 28-day mortality (Supplementary Fig. 2). There was no association between the multimodal vasopressor timing and the 28-day mortality among patients with SOFA of 8–10 scores. In the second model, the interaction between multimodal vasopressor groups and SOFA were not statistically significant (Supplementary Table 3).

4. Discussion

Septic shock is defined as persistent sepsis-induced hypotension despite adequate fluid resuscitation. Optimizing fluid resuscitation is the first step before initiation of vasopressor support, according to guidelines. Sepsis patients with a positive fluid balance have a higher mortality rate [19]. There is a growing body of evidence to recommend early initiation of norepinephrine therapy [20, 21], which is significantly associated with a lower amount of resuscitation fluids, less fluid accumulation, lower incidences of cardiogenic pulmonary edema and new-onset arrhythmia, increased shock control and improved mortality [20, 21, 22]. For patients with refractory shock, increasingly high doses of norepinephrine are needed to maintain the MAP target. Although high doses of norepinephrine therapy are associated with worse clinical outcomes, there is no high-quality evidence to recommend the use of other vasopressors over norepinephrine. Vasopressin is recommended as a secondary vasopressor for septic shock. Compared with norepinephrine, low-dose vasopressin did not reduce mortality rates in patients with septic shock in the VASST trial [12]. Vasopressin, however, significantly reduced the use of renal replacement therapy and improved renal function more than norepinephrine [23, 24, 25].

However, the optimal timing of administration of the secondary agent remains unclear. Recent retrospective study reported that arginine vasopressin initiation at a norepinephrine equivalent of >40 μg/min was associated with non-responsiveness to arginine vasopressin, which increased the risk of death in septic shock patients [26]. An observational study also demonstrated that the risk of in-hospital mortality increased by 20.7% for every 10 μg/min increase in the norepinephrine-equivalent dose up to 60 μg/min at the time of vasopressin initiation, but no association was observed when norepinephrine-equivalent dose exceeded 60 μg/min [13]. These findings suggest that the initiation of multimodal vasopressor therapy is associated with the improved clinical outcomes. Thus, similar to the use of broad-spectrum antimicrobials in sepsis, an early multimodal vasopressor strategy, also termed “broad-spectrum vasopressors” was proposed recently [2, 11]. Norepinephrine is the recommended first-line vasoactive drug, whereas epinephrine, phenylephrine dopamine, and vasopressin are usually considered second-line agents [3]. Despite being catecholamines, norepinephrine, epinephrine, phenylephrine and have different receptor activities [27]. From this perspective, using a norepinephrine and another type of catecholamine as a second-line vasopressor might also be “multimodal”. This study categorized the initiation of the second agent based on the norepinephrine dose. In septic shock patients receiving a maximum norepinephrine-equivalent dose >0.2 μg/kg/min, the earlier the multimodal vasopressors are initiated, the better prognosis, regardless of the agent type (catecholamine or vasopressin). Since high-dose norepinephrine therapy often results in adverse effects and is associated with poor outcomes in septic shock [28, 29], norepinephrine-sparing approaches are praised [11, 30]. This study found the EMMVT group received a significantly lower norepinephrine dose, indicating that early and broad-spectrum vasopressors reduced norepinephrine use, thus preventing toxic side effects from high-dose norepinephrine. In addition, it seemed that the usage of more types of vasopressors was associated with higher mortality (Fig. 2). Possibly, the worse hemodynamic instability facilitates the use of more vasopressors, resulting in a higher mortality.

In this study, the interaction between illness severity and multimodal vasopressor timing was also analyzed in the multivariable models. However, the interaction was not significant, indicating that the early multimodal vasopressor strategy was beneficial for patients with septic shock, regardless of illness severity. According to Guerci et al. [31], adding an early adjunct vasopressor to norepinephrine might not be necessary for “controlled shock”, in contrast to refractory shock. Future studies should also focus on efforts to individualize the use of vasopressors, considering the patient’s pathophysiological characteristics.

This study has some limitations. First, given the single-center retrospective nature of the study, the results obtained herein need to be confirmed in a well-designed prospective multicenter cohort before any extrapolation can be made. Second, although we demonstrated that early multimodal vasopressor therapy was beneficial for patients with septic shock, we failed to determine which vasopressor was the optimal second agent. Third, we failed to assess the timing of angiotensin II on septic shock, since no record about angiotensin II were searched in MIMIC database. Lastly, because of its retrospective nature, we failed to assess the impact of multimodal vasopressor therapy on long-term survival, as part of the patents in the MIMIC III database had date records of death only up to 90 days in the future.

5. Conclusions

For septic shock patients receiving norepinephrine as the first-line vasopressor and reaching a maximal norepinephrine-equivalent dose >0.2 μg/kg/min, early multimodal vasopressor therapy was associated with improved 28-day mortality, regardless of the illness severity. Given the limitations of the present retrospective study, randomized trials will be needed to conclusively endorse early multimodal therapy.

Availability of data and materials

The data presented in this study are available on reasonable request from the corresponding author.

Author contributions

TTW—conceptualization and writing–original draft. XHZ—data curation, formal analysis and methodology. PHY—conceptualization, supervision and writing–review & editing. JQZ—methodology and software. QL—project administration and writing–review & editing. XLY—Methodology, software and visualization. All authors approved the final version to be submitted for publication.

Ethics approval and consent to participate

Access to the MIMIC database was approved by the Institutional Review Boards of the Massachusetts Institute of Technology (Cambridge, MA, USA) and the Beth Israel Deaconess Medical Center (NO. 2001P001699). Informed consent was waived, because all patients were de-identified and all dates in the database were shifted to protect patient confidentiality.

Acknowledgment

Not applicable.

Funding

This work was supported by the Medicine and Health Technology Plan Project of Zhejiang Province (No. 2023KY809) and the Project of the Zhejiang Bureau of Traditional Chinese Medicine (No. 2022ZB375).

Conflict of interest

The authors declare no conflict of interest.

Supplementary material

Supplementary material associated with this article can be found, in the online version, at https://oss.signavitae.com/mre-signavitae/article/1831947679327764480/attachment/Supplementary%20material.docx.