Effects of very early start of norepinephrine in patients with septic shock: a propensity score-based analysis

Adult patients > 18 years or older fulfilling the diagnostic criteria for septic shock stated in the Surviving Sepsis Campaign: International Guidelines for Management of Severe Sepsis and Septic Shock: 2012 [37] and based on the previous 2001 SCCM/ESICM/ACCP/ATS/SIS International Sepsis Definitions Conference [38] were prospectively collected between January 2015 and February 2017 in one mixed-ICU in a university hospital in Colombia (Fundación Valle del Lili, Cali, Colombia). This original definition was maintained as inclusion criteria since it was valid during the period in which the database was constructed. The ethical and research committee involving human beings approved the use of the data (Protocol number 1238, IRB/EC approval number 099-2018, Fundación Valle del Lili, Cali, Colombia). The presence of infection was determined according to the Centers for Diseases Control and Prevention criteria [39]. For analysis purposes, however, septic shock was re-classified according to the current Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis 3.0), which consider the presence of suspected infection accompanying organ dysfunction, the use of vasopressors, MAP < 65 mmHg, and lactate levels > 2 mmol/L [40]. Meanwhile, patients with sepsis and vasopressor requirement but without hyperlactatemia were re-classified and analyzed as sepsis-related acute circulatory failure. Surviving patients requiring less than 6 h of vasopressor support were not included in the initial main analysis, as they could not be representative of septic shock. Nevertheless, these patients were also evaluated in an ulterior sensitivity analysis. Patients < 18 years old, pregnant women, patients with liver failure (prothrombin time > 15 s or international normalized ratio ≥ 1.5 and any hepatic encephalopathy), advanced liver cirrhosis (Child-Pugh C), and those with do-not-resuscitate orders were excluded.

Very early start of vasopressor (VE-VPs) was defined as that vasopressor support initiated within the next hour or even before the first fluid load with resuscitative intention (FRLoad). Those patients in whom vasopressor support was started > 1 h after the FRLoad were classified as delayed VPs (D-VPs). In each case, the start of vasopressor (VPs) was identified and used as a reference point to determine the time elapsed from the first hypotension episode (FHypo) and from the FRLoad. The decision to start vasopressor support was always taken by the attending physician according to his clinical judgment. The delay time until the start of antibiotics was also recorded with respect to the first hypotension episode. In addition, time intervals from FHypo, FRLoad, and VPs up to ICU admission were also recorded. The volume of resuscitation fluids received before VPs was also registered. Then, the volume of resuscitation fluids and dose of vasopressors were recorded at 2-h intervals from the VPs up to 8 h, and then, 12, 18, and 24 h after VPs. Net fluid balance was also recorded at FHypo, FRLoad, VPs, and also 8 h (8H) and 24 h (24H) after the start of vasopressor support.

General demographics including age, APACHE II, comorbidities, and origin of the patient (emergency room, general ward, or intensive care unit) were registered. Heart rate and arterial pressure were also recorded at FHypo, FRLoad, VPs, and at 2, 4, 6, 8, and 24 h after the VPs. Multiple organ dysfunction was assessed using the Sequential Organ Failure Assessment Score (SOFA) [41]. Ventilator-free days, requirement of renal replacement therapy (RRT), and RRT-free days were also calculated. Finally, ICU and hospital length of stay were recorded along with UCI, in-hospital, and 28 days of mortality.

Patients followed an early quantitative resuscitation protocol aiming to target: (a) MAP ≥ 65 mmHg; (b) urine output > 0.5 mL/kg/min; (c) ScvO₂ ≥ 70%, when available; and (d) normalization of lactate levels or decreasing of 20% each 2 h in lactate levels. Fluid resuscitation was performed administering repeated fluid challenges with crystalloids and/or albumin 4%, using the central venous pressure (CVP) as a dynamic safety limit during fluid loads in patients with a central line in place. Hydroxyethyl starches (HES) were not used. The usual protocol in our institution includes the use of pulse pressure and stroke volume variations to guide fluid resuscitation (when usable). Additionally, echocardiographic determination of velocity-time integral (VTI) before and after passive leg raising (PLR), and end-expiratory occlusion maneuvers were used whenever applicable. The clinical assessment of peripheral perfusion (e.g., measuring capillary refill time [42] and/or the evaluation of mottling score [43]) and the use of advanced monitoring of cardiac output were allowed at the discretion of the attending physician. Norepinephrine was the first-choice vasopressor used to achieve MAP ≥ 65, while vasopressin titrated up to 0.04 UI/min was also allowed to increase MAP or to decrease norepinephrine dose, but never as a single vasopressor. Dobutamine up to 20 μcg/kg min was used in case of myocardial dysfunction, when ScvO₂ or lactate goals were not achieved or when clinical signs of hypoperfusion persisted despite adequate fluid resuscitation. Mechanical ventilation was used when indicated, providing light sedation (midazolam or propofol) and analgesia (fentanyl). Red blood cell transfusion was used to maintain hemoglobin levels at or above 7.0 g/dl or > 10.0 g/dl in case of cardiac ischemia. Low-dose hydrocortisone was used when the vasopressor requirement did not decrease during the first 6 h of resuscitation in the presence of an adequate intravascular volume. Glycemic control was adjusted to maintain glucose levels < 150 mg/dL, while thrombosis prophylaxis and stress ulcer protection were also provided according to international guidelines valid at the time in which patients were treated [37].

Patients meeting eligibility criteria and subjected to VE-VPs were propensity-matched with those subjected to D-VPs. For that, factors potentially influencing the decision of very early vasopressor support such as source of admission (emergency room, intensive care unit), age, chronic, and acute comorbidities (hypertension, coronary disease, chronic heart failure, end-stage renal failure, chronic atrial fibrillation, chronic use of steroids, previous stroke, diabetes, cancer, chronic obstructive pulmonary disease, cirrhosis Child-Pugh C, acute myocardial infarction, acute heart failure, acute stroke, acute atrial fibrillation), diastolic blood pressure, systolic blood pressure and heart rate at the VPs, the heart rate/diastolic blood pressure ratio at the FRLoad, and arterial lactate levels at the VPs were included in a logistical model to estimate the propensity scores. After fitting the propensity score, a nearest neighbor-matching algorithm extracted 1:1 matched pairs of VE-VPs and D-VPs individuals. The effect of early start of vasopressors on mortality at day 28 was assessed using a Cox-proportional hazards model adjusted by SOFA score at day 1, the presence of hyperlactatemia (septic shock according to Sepsis 3.0 definition), delay time of antibiotic administration, and the net fluid balance at 24 h. In addition, adjunctive therapies (e.g., renal replacement therapies, vasopressin, and steroid use) were also used as covariables. A conditional forward stepwise technique was used after verifying all subsets selection, while the proportional hazards assumption was tested on the basis of Schoenfeld residuals.

Repeated measures ANOVA were used to evaluate the time-course of vasopressor dose and cumulated resuscitation fluids during the first 8 h of resuscitation and the inter-group differences between VE-VPs and D-VPs.