We performed a retrospective cohort study based on a single-center, prospectively collected registry. The emergency department (ED) of Asan Medical Center has participated the Korean Shock Society (KoSS) septic shock registry study since October 2015. The KoSS septic shock registry study collects data pertaining to patients with septic shock at 10 teaching hospitals across South Korea [10]. Following the protocol of KoSS septic shock registry study, our center included all adult patients (≥ 19 years) when they showed evidence of refractory hypotension or hypoperfusion with suspected or confirmed infection within 6 hours after ED presentation [1,11]. Refractory hypotension was defined as sustained hypotension despite an intravenous fluid supply of ≥ 20 to 30 mL/kg or the need for vasopressors to preserve an SBP of > 90 mmHg or MAP of > 70 mmHg [11]. Hypoperfusion was defined as a lactate concentration of > 4 mmol/L [12]. Patients were excluded if they refused intensive management or consented to a “do not resuscitate” order, developed septic shock 6 hours after ED arrival, were directly transferred to another hospital at ED, or refused enrollment in the septic shock registry [13-15]. Among the septic shock patients, about 75% of patients were finally enrolled. The most common reason for exclusion was having “do not resuscitate” order (56%), followed by refusal of enrollment (28%), development of septic shock 6 hours after ED arrival (12%), and a direct transfer to another hospital at ED (4%). Patients were followed up until 28 days after hospital admission or the time of death. The study protocol was approved by the Institutional Review Board of Asan Medical Center (IRB No. 2018-1020), and informed consent was obtained prior to patient enrollment.

In this study, we extracted the patients who met the criteria of the Third International Consensus for Sepsis and Septic Shock (Sepsis-3), and admitted to ED between June 1, 2016, and April 30, 2018. The Sepsis-3 criteria include persistent hypotension requiring vasopressors and a serum lactate concentration of > 2 mmol/L after adequate fluid administration [1]. Patients were categorized into 28-day survivor and non-survivor groups.

All the patients were treated with a protocol-driven resuscitation bundle therapy in accordance with the Surviving Sepsis Campaign guidelines, which included adequate fluid resuscitation (at least 30 mg/kg of crystalloids), vasopressor administration, acquisition of blood cultures before starting the antimicrobial therapy and antibiotic treatment [2,11]. We used norepinephrine (NE) as an initial vasopressor agent and added vasopressin at a dose of 0.04 units/minute when vital signs were unstable despite the use of NE at a dose of ≥ 0.2 µg/kg/min. And we added epinephrine when vital sign was not stabilized, if NE infusion rate of 0.4 µg/kg/min or more was required. Arterial and central venous lines were routinely inserted and monitored continuously in patients who required high-dose vasopressors, defined as a dosage of ≥ 0.1 mg/kg/min, to achieve MAPs of ≥ 65 mmHg.

Information on demographic and clinical characteristics, including age, sex, comorbid diseases, infection focus, laboratory findings, and 28-day mortality were obtained from the registry. We analyzed the vital signs measured during the first 6 hours after the recognition of septic shock. These data were extracted from the electronic medical records. SBP, DBP, and heart rate (HR) were recorded automatically every minute in the intensive care unit and manually at least once every hour in the ED. As manual recordings input into the electronic medical records in the ED may contain errors, SBPs higher than 220 mmHg or those lower than their corresponding DBP (or 20 mmHg) were discarded because such values are improbable.

TWA hemodynamic parameters were calculated, including SBP, DBP, HR, and MAP. The MAP was calculated as 2 / 3 × DBP + 1 / 3 × SBP. Because of their unequal intervals of measurements, TWA was used to estimate their averages. TWA parameters were calculated using the following equation (Fig. 1).

In addition, the TWA of PP (SBP − DBP) and SI (HR / SBP) were determined. The primary outcome was 28-day mortality.

Non-normally distributed continuous variables, as determined using Kolmogorov-Smirnov tests, are reported as median and interquartile ranges (IQRs) and compared using Mann-Whitney U tests. Categorical variables are reported as number and percentage and compared using the chi-square or Fisher exact test, as appropriate. The associations of the baseline characteristics and hemodynamic parameters with the 28-day mortality rates were analyzed using univariable logistic regression. The variables with an entry-level significance of p values of < 0.10 in univariable analysis, age, and sex were selected for multivariable stepwise logistic regression analysis with a backward elimination method. The results were summarized by estimating the adjusted odds ratios (ORs) and 95% confidence intervals (CIs). Models were tested for goodness-of-fit using the Hosmer-Lemeshow method (Supplementary Table 1). Two-sided p values of < 0.05 were considered statistically significant. All statistical analyses were performed using IBM SPSS Statistics version 21.0 for Windows (IBM Corp., Armonk, NY, USA).

During the study period, 523 patients with septic shock were enrolled in the septic shock registry, among whom 340 fulfilled the clinical criteria for septic shock according to the Sepsis-3 definitions and were included in this study (Fig. 2). The median number of vital signs checks per patient enrolled with septic shock was 10 (IQR, 8 to 14), and only one patient died within 6 hours of presentation. Among the 340 patients, 92 (27.1%) died within 28 days after presentation to the ED.

The 340 patients included 215 men (63.2%) and 125 women (37.8%) with a median age of 66.0 years (Table 1). Comorbid diseases did not differ significantly between the 28-day survivors and non-survivors. The respiratory tract (n = 109, 32.1%) was the most common focus of infection, followed by the hepatobiliary tract (n = 106, 31.2%) and the urinary tract (n = 46, 13.5%). Respiratory tract infection was significantly more frequent (44.6% vs. 27.4%, p = 0.001), whereas urinary tract infection was significantly less frequent (5.4% vs. 16.5%, p = 0.003) in the 28-day non-survivors than in the survivors.

Assessments of laboratory findings revealed that prothrombin time (international normalized ratio, 1.4 vs. 1.3; p = 0.005), lactate concentration (6.5 mmol/L vs. 4.8 mmol/L, p = 0.001), C-reactive protein concentration (15.9 mg/dL vs. 11.5 mg/dL), and sequential organ failure assessment (SOFA) score (7.0 vs. 6.0, p = 0.001) were significantly higher in the non-survivors than in the survivors, whereas white blood cell and platelet counts did not significantly differ between the two groups.

Table 2 shows the TWA hemodynamic parameters of the patients with septic shock, grouped by 28-day survival. The median TWP-PP was significantly higher in the non-survivors (43.0 vs. 39.2, p = 0.020), whereas other hemodynamic parameters, including MAP, SBP, DBP, and SI, did not significantly differ between the two groups.

The multivariable analysis revealed that 6-hour TWA PP (adjusted OR, 1.28 mmHg; 95% CI, 1.012 to 1.622; p = 0.039), lung infection (adjusted OR, 2.05; 95% CI, 1.209 to 3.470; p = 0.008), C-reactive protein concentration (adjusted OR, 1.03; 95% CI, 1.004 to 1.049; p = 0.021), lactate concentration (adjusted OR, 1.14; 95% CI, 1.057 to 1.234; p = 0.001), and SOFA score (adjusted OR, 1.10; 95% CI, 1.014 to 1.192; p = 0.022) were independent predictors of 28-day mortality in patients with septic shock (Table 3).

When TWA PP was divided into quartiles (< 34, 34 to 40, 40 to 48, and > 48 mmHg), the 28-day mortality rates in these subgroups were 22.4%, 22.4%, 22.4%, and 41.2%, respectively (Fig. 3). The multivariable logistic regression analysis revealed that PPs of > 48 mmHg (OR, 2.25; 95% CI, 1.272 to 3.981; p = 0.005) were independently associated with the 28-day mortality rate.