a b s t r a c t

Study objective: Quantify the correlation between Blood pressure variability (BPV) and markers of illness severity: serum Lactate or Sequential Organ Failure Assessment scores.

Methods: We performed a secondary analysis of data from a prospective, observational study evaluating fluid resuscitation on adult, septic, ED patients. Vital signs and fluid infusion volumes were recorded every 15 min during the 3 h following ED arrival. BPV was assessed via average real variability (ARV): the average of the absolute differences between consecutive BP measurements. ARV was calculated for the time periods before and after 3 fluid infusion milestones: 10-, 20-, and 30-mL/kg total body weight (TBW). Spearman’s rho correla- tion coefficient analysis was utilized. A p-value b 0.05 was considered statistically significant.

Results: Forty patients were included. Mean fluid infusion was 33.7 mL/kg TBW (SD 22.1). All patients received fluid infusion >= 10 mL/kg TBW, 25 patients received fluid infusion N 20 mL/kg TBW, and 16 patients received fluid infusion N 30 mL/kg TBW. Mean initial LAC was 4.0 mmol/L (SD 3.2). Mean repeat LAC was 3.1 mmol/L (SD 3.2), obtained an average of 6.6 h (SD 5.3) later. Mean SOFA score was 7.0 (SD 4.4). BPV correlated with both follow-up LAC (r = 0.564; p = 0.023) and SOFA score (r = 0.544; p = 0.024) among the cohort that received a fluid infusion N 20-mL/kg TBW. Conclusion: With the finding of a positive correlation between BPV and markers of illness severity (LAC and SOFA scores), this pilot study introduces BPV analysis as a real-time, non-invasive tool for continuous sepsis monitoring in the ED.

(C) 2017

Introduction

Sepsis is the 11th leading cause of death in the United States [1-2]. Early detection and treatment of sepsis can significantly improve clini- cal outcomes [3-4]. Because a large proportion of septic patients present

Abbreviations: ARV, average real variability, the average of the absolute differences between consecutive measurements within a series (mm Hg for BPV, beats per minute for HRV); BP, blood pressure (mm Hg); BPV, blood pressure variability, the change in blood pressure over time; HRV, Heart rate variability, the change in heart rate over time; LAC, serum lactate (mmol/L).

? Funding: Funding for this study was provided by Wayne State University School of

Medicine and Vidacare/Teleflex Corporation.

?? Conflicts of interests: There are no conflicts of interest for the primary author or any of

the co-authors.

? Prior abstract publication/presentation: This research has been presented (abstract)

at the 2017 meeting for the Society of Critical Care Medicine, Honolulu, Hawaii, on January 23, 2017.

* Corresponding author at: Wayne State University School of Medicine, 540 E. Canfield St., Detroit, MI 48201, USA.

E-mail address: [email protected] (J.E. Nouriel).

initially to the emergency department (ED), ED providers have a unique opportunity to alter the course of disease [5]. However, identifying sep- sis and its severity at the time of ED arrival can be challenging. Rapid, non-invasive tools for monitoring early sepsis management are needed. Blood pressure control is an essential component of sepsis treatment. The 2016 Surviving Sepsis Campaign guidelines recommend maintaining a mean arterial pressure (MAP) >= 65 mm Hg throughout the first 6 h of resuscitation [6]. Additionally, recent defini- tions of sepsis have placed a great deal of emphasis on the patient’s pre- senting BP, categorizing those patients with a persistent SBP b 90 mm Hg and Serum lactate level N 2 mmol/L despite adequate fluid resuscita- tion as having septic shock [7]. Septic shock represents an advanced state of sepsis with increased mortality [8]. Therefore, blood pressure measurement may serve a prognostic role in the evaluation of sepsis

and can help to guide therapy.

The majority of previous studies evaluating the prognostic value of BP among septic patients have focused on determining the optimal MAP [9]. However, BP is sensitive to neuronal and hormonal systems, and oscillates in both short- (minutes to hours) and long-term (days

https://doi.org/10.1016/j.ajem.2017.09.017

0735-6757/(C) 2017

to months) periods [10]. Blood pressure variability (BPV) is a measure of this oscillation of BP over time. High BPV has been shown to be indepen- dently associated with and prognostic for greater levels of multi-organ damage (MOD) [11-14]. Hence, it is reasonable to expect that septic patients, who by definition suffer MOD, would also have high BPV. In a study of septic patients within the first 48 h of hospital admission, Pandey et al. found that a greater severity of illness was associated with statistically significant greater systolic and Diastolic BPV [15]. In contrast, Berg et al. found that BP oscillation occurred at reduced frequencies in mechanically-ventilatED septic patients, as compared to healthy, spontaneously breathing patients [16]. Thus, it is not clear whether septic patients suffer from labile BP.

Characterization of the relationship between BPV and “traditional” makers of sepsis severity (e.g., lactate, Sequential Organ Failure Assess- ment (SOFA) scores) could provide clinicians with a non-invasive tool to monitor sepsis management. Although tissue hypoperfusion is only one possible mechanism for lactemia, high serum lactate and low Lactate clearance values are associated with worse clinical outcomes among septic patients [17-20]. Current SSC guidelines recommend that resuscitation should aim to normalize lactate (to a level b 1 mmol/L) as rapidly as possible [6]. However, serial lactate measurement represents static indicators of perfusion, and occurs over a period of many

resuscitation: fluid administration start time, fluid administration con- clusion time, and volume of fluid infused during each 15-minute inter- val. Patients were followed from the time of ED arrival, but patient identifiers were only paired with the anonymously-collected data if the patient or their legally-authorized representative consented to study inclusion. Patients who did not consent to enrollment had their data used anonymously, without patient identifiers. Data collected fol- lowing obtainment of informed consent included: comorbidities, results of all laboratory and diagnostic tests during the hospitalization, pharma- cologic therapy, and discharge diagnoses at time of both ED disposition and hospital discharge. At the time of enrollment, study participants were assigned a unique Study Number. All HIPAA regulations were strictly followed in order to maintain the privacy of study participants. Extracted data was compiled in a Microsoft(C) Excel spreadsheet, maintained on a password-protected institutional webserver. For this subgroup analysis, only consented patient charts were included.

We assessed systolic BPV via a previously established statistical tool for BPV, average real variability (ARV) which is calculated as the average of the absolute differences between consecutive BP measurements within a series [24]:

1 NX-1

hours, which may not be able to provide clinicians with adequate “real- time” feedback on the efficacy of their care for septic patients. While the SOFA score evaluates septic patients for the presence of MOD, thereby

ARV = N-1

k=1

| BP_k+1-BP_k |

assisting clinicians in estimating illness severity and mortality risk, these scores are not generally available early in the patient’s resuscitation [21]. Consequently, it is necessary to develop additional tools that would be available to the emergency physician in “real-time” during the first few hours following ED arrival. Such tools are likely to help lower the high mor- tality rate associated with sepsis [22-23].

Previous data characterize some of these relationships, but only at a more advanced stage in the patient’s illness. Our study sought to evalu- ate BPV at an earlier time point in the course of the disease – within the first three hours following ED arrival. Our objectives were to quantify both the correlation between BPV and Serum lactate levels and the cor- relation between BPV and SOFA scores. We hypothesized that a positive relationship exists between BPV and these established markers of sepsis severity.

Methods

We conducted a secondary analysis of data obtained from an on- going multi-center prospective, observational study, Shock Access For Emergent Resuscitation (SAFER). SAFER is designed to evaluate the relationship between early fluid resuscitation and clinical outcomes in patients with undifferentiated hypotension (SBP b 90 mm Hg). Our sub-analysis of SAFER data includes only septic patients, retrospectively identified by the presence of hypotension on ED arrival with a positive laboratory or other diagnostic test finding of bacterial infection. All patients were diagnosed with sepsis by the admitting physician, and all had documented sources of bacterial infection. All patients were en- rolled from December 2014 to August 2016 at Sinai-Grace Hospital and Detroit Receiving Hospital (Detroit, MI). Both are large, Tertiary care hospitals staffed by emergency medicine resident physicians supervised by emergency medicine attending physicians. This protocol was approved by the Institutional Review Board at Wayne State University (WSU), the affiliated academic institution for both hospitals.

Patients were eligible for enrollment if they were at least 18 years

old, presented with a SBP b 90 mm Hg, and had intravascular volume– depletion requiring fluid bolus administration of at least 1 L of IV fluid within the first hour following ED arrival (per discretion of treating ED physician). Patient vital signs (BP, HR, RR, temperature) were recorded initially at presentation and again at 15 min intervals throughout the first 3 h following ED arrival. The relevant SAFER data for our secondary analysis included details on blood, crystalloid and colloid fluid

ARV was calculated for systolic blood pressure (SBP) measurements during the time periods before and after three milestones of intravenous fluid infusion: 10 mL/kg of total body weight (TBW), 20 mL/kg TBW, and 30 mL/kg TBW. (N) represents the number of SBP measurements within a given series. Hemodynamic data was not used once vasopres- sor therapy was initiated, due to the direct alteration on vasomotor activity. ARV was additionally utilized to assess heart rate oscillation.

SOFA scores were calculated for all patients and were calculated from the worst values for the indicated parameters during the initial 24 h following ED arrival. When patients lacked a measurement of PaO₂, the respiratory component of SOFA scores was calculated via a previously validated method that utilized pulse oximetry (SpO2/FiO2 ratio) in lieu of an arterial blood gas analysis (PaO2/FiO2 ratio) [25-26]. Bilirubin data was not available for 15 patients and normal bilirubin values were imputed for the SOFA score calculation. Serum lac- tate measurements included both an initial specimen and the next sub- sequent specimen sampled within 24 h of the initial specimen. Lactate clearance was calculated as the percentage of the difference between ini- tial and final lactate divided by initial lactate. A positive value indicated a decrease in serum lactate, while a negative value indicated an increase in serum lactate. Spearman’s rank-order correlation coefficient analysis was utilized to calculate the association between a hemodynamic variable (BPV or HRV) and a marker for illness severity (serum lactate or SOFA scores). Any p-value b 0.05 was considered statistically significant. Statistical analysis of the data was performed using IBM SPSS Statistics, Version 23 (Armonk, NY, USA).

Results

Forty patients from among the 181 consented SAFER patients quali- fied for inclusion in this sub-analysis. Fig. 1 is a flow chart for the protocol of patient enrollment within the SAFER study (A) and within the BPV sub-analysis (B). Table 1 provides a patient summary.

For patients who received a fluid infusion of 10 mL/kg TBW, the average BPV before and after the 10 mL/kg TBW fluid infusion was

13.6 mm Hg and 10.4 mm Hg, respectively. Three patients with subject numbers of 28, 40, and 27 were outliers for the BPV after receiving 10 mL/kg TBW fluid infusion, demonstrated by BPV values equal to

24.6 mm Hg, 26.3 mm Hg, and 26.6 mm Hg, respectively. For patients

who received a fluid infusion of 20 mL/kg TBW, the average BPV before and after the 20 mL/kg TBW fluid infusion was 13.7 mm Hg and

Fig. 1. (A) Flowchart of SAFER enrollment protocol; (B): Flow chart of BPV sub-analysis enrollment protocol.

10.2 mm Hg, respectively. Patient with subject number 27 was an outli- er for BPV before receiving 20 mL/kg TBW fluid infusion, with a BPV value of 33.8 mm Hg. Patients with subject numbers 28 and 40 were

Table 1

Patient summary

Parameter	Value	SD
Mean age (years)	59	15
Female (%)	58	–
Mean TBW (kg)	78	38
Mean fluid administration mL/kg TBW	33.7	22.1
Number of patients who received fluid administration 10 mL/kg TBW	40	–
Number of patients who received fluid administration 20 mL/kg TBW	25	–
Number of patients who received fluid administration 30 mL/kg TBW	16	–
Mean initial lactate mmol/L (n = 39)	4	3.2
Mean subsequent lactate mmol/L (n = 31)	3.1	3.2
Mean lactate clearance (%)	20	58
Mean time between initial and subsequent lactate (hours)	6.6	5.3
Sequential Organ Failure Assessment score (n = 40)	7.0	4.4

TBW: total body weight.

outliers for BPV after receiving 20 mL/kg TBW fluid infusion, with a BPV value equal to 24.7 mm Hg. For patients who received a fluid infusion of 30 mL/kg TBW, the average BPV before and after the 30 mL/kg TBW fluid infusion milestone was 15.5 mm Hg and 12 mm Hg, respectively. See Fig. 2. Table 2 presents the corresponding BPV values for Fig. 2.

Of the 25 patients who received a Fluid replacement of 20 mL/kg TBW, an ARV calculation for BPV following a fluid administration of 20 mL/kg TBW was possible for 17 patients (because the ARV calcula- tion requires 2 data points). All 17 patients, who both received 20 mL/kg TBW fluid infusion and had a post-fluid infusion ARV calcula- tion also had an initial lactate measurement versus 16 out of these 17 patients had a subsequent lactate measurement. BPV during the time period following a fluid infusion N 20 mL/kg TBW positively correlated with follow-up lactate levels (r = 0.564; p = 0.023). BPV following a fluid infusion N 20 mL/kg TBW positively correlated with SOFA scores (r = 0.544; p = 0.024). The correlation between BPV following a fluid infusion N 20 mL/kg TBW and initial lactate levels approached statistical significance (r = 0.474; p = 0.055). Out of the 16 patients who received a fluid infusion of 30 mL/kg TBW, BPV prior to achieving this fluid

Fig. 2. Boxplot reflecting the interquartile range (25th to 75th percentile range) for the variability in systolic BPV determined by ARV. Variations were determined for SBP measurements before and after 3 grades of fluid infusion milestones 10-, 20-, and 30-mL/kg TBW. The solid, dark horizontal line in each box represents median values (50th percentile). Patients with subject numbers of 27, 28, and 40 were outliers.

milestone positively correlated with SOFA scores (r = 0.794, p = 0.000). Out of the 16 patients who received a fluid infusion of 30 mL/kg TBW, 15 patients had a subsequent lactate measurement. For these 15 patients, the positive correlation between BPV prior to the 30 mL/kg TBW fluid milestone and subsequent lactate approached statistical significance (r = 0.488, p = 0.065). 36 patients had both an ARV calculation for BPV following a fluid infusion of 10 mL/kg TBW and an initial lactate measurement. For these 36 patients, the correla- tion between BPV following fluid infusion of 10 mL/kg TBW and initial lactate measurements was not statistically significant (r = 0.288, p = 0.088). See Table 3.

Out of the 40 patients who received a fluid infusion of 10 mL/kg TBW, an HRV calculation was possible for 37 patients for the time period after achieving this fluid milestone. Of these 37 patients, 36 patients had an initial lactate versus 29 patients had a subsequent lactate. heart rate variability following a fluid infusion of 10 mL/kg TBW correlated with both initial (r = 0.345; p = 0.039) and follow-up lactate (r = 0.419; p = 0.024). The correlation between HRV following 10 mL/kg TBW and SOFA score was not statistically significant (r = 0.281; p = 0.092). Out of the 25 patients who received a 20 mL/kg TBW fluid infu- sion, an ARV calculation for HRV prior to achieving this fluid milestone

was possible for 24 patients. HRV prior to receiving a fluid infusion of 20 mL/kg TBW positively correlated with SOFA scores (r = 0.456; p = 0.025). The correlation between HRV prior to receiving a fluid infusion of 20 mL/kg TBW and initial lactate approached statistical significance (r = 0.386; p = 0.069). Out of the 40 patients who received a fluid in- fusion of 10 mL/kg TBW, an ARV calculation for HRV prior to receiving fluid infusion 10 mL/kg TBW was possible for 36 patients. Of these 36 patients, 35 patients had an initial lactate versus 28 had an additional subsequent lactate. Both the correlation between HRV prior to receiving 10 mL/kg TBW fluid infusion and initial lactate (r = 0.298; p = 0.0.82) and the correlation between HRV prior to receiving 10 mL/kg TBW fluid infusion and subsequent lactate were not statistically significant (r = 0.328; p = 0.089). See Table 4.

Discussion

To the best of our knowledge, this is the first study designed to ex- amine the clinical utility of BPV in monitoring the progression of early sepsis management in the ED. We assessed BPV via ARV, a previously established statistical tool for quantifying BPV [24]. Within the context of assessing the prognostic significance of BPV for cardiovascular events

Table 2

Interquartile ranges for BPV over the course of fluid resuscitation.

Patient cohort based on fluid resuscitation status	BPV (mm Hg) of 25th percentile	BPV (mm Hg) of 50th percentile	BPV (mm Hg) of 75th percentile	Difference between BPV of 75th percentile and BPV of 25th percentile
Before receiving 10 mL/kg TBW	6.5	11.5	19.5	13.0
After receiving 10 mL/kg TBW	6.4	9.1	13.4	7.0
Before receiving 20 mL/kg TBW	8.1	12.3	16.9	8.8
After receiving 20 mL/kg TBW	5.7	7	12.5	6.8
Before receiving 30 mL/kg TBW	9.2	15.3	19.6	10.4
After receiving 30 mL/kg TBW	4.3	7.4	17.7	13.4

Table 3

Correlation coefficients between blood pressure variability (BPV) and markers of illness severity.

Patient cohort Correlation coefficients between BPV (mm Hg) and initial lactate (mmol/L)

Correlation coefficients between BPV

(mm Hg) and subsequent lactate (mmol/L)

Correlation coefficient between BPV (mm Hg) and SOFA score

Post fluid administration of 10 mL/kg TBW 0.288 (p = 0.088) – –

(n = 36)

Post fluid administration of 20 mL/kg TBW 0.474 (p = 0.055) 0.564 (p = 0.023) 0.544 (p = 0.024)

(n = 16)^a (n = 17)

Prior to fluid administration of 30 mL/kg TBW – 0.488 (p = 0.065) 0.794 (p = 0.000) (n = 15)^b n= 16

TBW: total body weight; SOFA: Sequential Organ Failure Assessment.

^a One patient lacked a subsequent lactate.

^b One patient lacked a subsequent lactate.

(coronary artery disease, stroke and congestive heart failure), Mena et al. developed ARV against standard deviation as a measure of variability and derived ARV from the total variability concept of real analysis in mathematics [27]. We found that both the average and median BPV values decreased after each grade of fluid resuscitation (10 mL/kg TBW, 20 mL/kg TBW, and 30 mL/kg TBW), indicating that patients on average experienced a more volatile blood pressure prior to clinical in- tervention. The interquartile ranges for BPV after a fluid resuscitation of 10- and 20-mL/kg TBW were less than the interquartile ranges of BPV pre-fluid resuscitation. In contrast, the interquartile range of BPV after a fluid resuscitation of 30 mL/kg TBW was greater than the interquartile range of BPV pre-fluid resuscitation. BPV following a fluid resuscitation of 20 mL/kg TBW demonstrated the smallest interquartile range while BPV following a fluid resuscitation of 30 mL/kg TBW demonstrated the largest interquartile range. 3 patients were consistent outliers from the interquartile range of BPV.

In order to assess the clinical significance of BPV, we investigated the

correlation between BPV and markers of illness severity (lactate and SOFA scores). In our analysis, serum lactate is the “gold standard” to which BPV is compared. Similar to the decrease in BPV identified follow- ing clinical intervention, mean initial lactate decreased from 4 mmol/L (SD 3.2) to 3.1 mmol/L (SD 3.2) for mean subsequent lactate, which was obtained an average of 6.6 h (SD 5.3) later. We found a positive correlation between BPV and serum lactate levels, suggesting that per- sistently increased BPV may indicate hyperlactemia. This propensity to predict hyperlactemia via BPV analysis promotes early clinical interven- tion, which has been shown to improve clinical outcomes [3-4].

Mean SOFA score was 7.0 (SD 4.4), which approximately represents a 15% mortality risk [21]. Our main secondary finding was that BPV pos- itively correlated with SOFA scores, which further supports our hypoth- esis that a positive relationship exists between BPV and illness severity. Jones et al. found that applying SOFA scores to patients with severe sep- sis with evidence of hypoperfusion at the time of ED presentation func- tioned with fair to good accuracy at predicting in-hospital mortality, which consequently suggests that SOFA scores at ED presentation is an acceptable method for risk stratification and prognosis determina- tion [28]. However, SOFA scores are not readily available during the patient’s early management in the ED versus BP measurement is an es- sential, non-invasive monitor of a patient’s hemodynamic status. Addi- tionally, calculation of a SOFA score over a short time interval (b 1 h) is not practical within the clinical setting and is unlikely to be altered over

such a short time period. Therefore, for a parameter that is readily avail- able at the patient’s bed-side, can be monitored for changes over short time intervals and is strongly correlated with and potentially predicts the patient’s SOFA score, the monitoring of such a parameter likely has a high degree of clinical utility for the management of the septic patients within the ED.

Sepsis is characterized as a complex state of systemic inflammation following an infection, with a resultant increase in inflammatory medi- ators (TNF-alpha) that elicits diffuse vasodilation [29]. Conflicting vasodilatory and vasoconstricting influences (lactic acidemia versus re- flex Sympathetic activity) may explain our finding that hyperlactemia positively correlated with labile SBP. Although inflammation appears to be the primary mechanism underlying the altered vasomotor tone seen in sepsis, other factors are being discovered. Sharshar et al. found an association between septic shock and apoptosis of neurons in the cardiovascular autonomic centers [30]. Autonomic imbalance with in- creased activity in the parasympathetic arm and an uncoupling of cardi- ac tissue to autonomic influence have also been proposed as explanations for the hemodynamic changes in sepsis [31-32]. It should be noted that the parasympathetic arm does not significantly innervate human vasculature. Clarifying this multifactorial regulation of BP will aid future clinical investigations of BPV.

The clinical significance of BPV may relate to autoregulation which normally maintains a constant blood flow when BP changes. Such mech- anisms are uniquely present for cerebral and renal blood flow [33-34]. Recent data suggest that sepsis represents a state of dysregulation in ce- rebral autoregulation [35-36]. Since Cerebral autoregulation mecha- nisms are arguably active during changes in BP, a highly volatile BP may stress impaired autoregulation mechanisms. This potentially intro- duces a deleterious effect. Acute kidney injury is a common prob- lem following sepsis [37]. Renal autoregulation also appears to be dysregulated during sepsis [38]. We speculate a theoretical harm from a volatile BP, particularly when autoregulation of renal perfusion is im- paired or has diminished functional capacity. According to our findings, BPV analysis provides a means to mitigate septic patients who are at risk for tissue hypoperfusion.

The small sample size was the primary limitation to our study. Fur- thermore, 15 out of 40 patients (38%) had all 13 systolic blood Pressure measurements (ED triage BP plus a BP every 15 min over the first 3 h of ED admission); 13 patients (33%) had 1 missing SBP measurement; 4

patients (10%) had 2 missing SBP measurements; and 4 patients (10%)

Table 4

Correlation coefficients between Heart rate variability and markers of illness severity

Patient cohort Correlation coefficient between HRV (bpm) and initial lactate (mmol/L)

Prior to fluid administration of 10 mL/kg TBW 0.298 (p = 0.082)

n= 35

Post fluid administration of 10 mL/kg TBW 0.345 (p = 0.039)

n= 36

Correlation coefficient between HRV (bpm) and subsequent lactate (mmol/L)

0.328 (p = 0.089) n = 28

0.419 (p = 0.024) n = 29

Correlation coefficient between HRV (bpm) and SOFA score

–

0.281 (p = 0.092) n= 37

Prior to fluid administration of 20 mL/kg TBW 0.386 (p = 0.069)

n= 23^a

– 0.456 (p = 0.025)

n= 24

^a One patient lacked an initial lactate.

had 3 missing SBP measurements; and 4 patients (10%) had N 3 missing SBP measurements. Additionally, our study lacked a control group. A planned larger study involving both septic and non-septic patients may provide stronger evidence on the relationship between BPV and ill- ness severity. Furthermore, blood pressure is influenced by many differ- ent factors, including time of day, posture, respiratory rate, pre-existing pathophysiology and home Pharmacologic therapy. Within a clinical setting, it is virtually impossible to account for all of these variables fully. Since patients with a history of chronic hypertension are in a state of increased basal sympathetic activity, future studies on BPV in sepsis patients may compare Pre-existing conditions that are likely to be hemodynamically significant to those that are less likely to impact patient hemodynamics. Home medications with hemodynamic signifi- cance will also need to be studied, including serum concentrations to fully understand the interplay between native hormonal regulators and exogenous substances.

We designated three milestones of fluid infusion with the maximum

milestone equal to 30 mL/kg TBW, based upon the 2016 SSC guidelines recommendation of a fluid resuscitation of 30 mL/kg TBW within the first 3 h of ED admission [6]. We added the 10 mL/kg and 20 mL/kg fluid administration milestones in order to improve the granularity of the data, and because preliminary data suggested that not all patients would receive the requisite 30 mL/kg TBW fluid infusion during the 3- hour study period [39-40]. As such, less than half of our included pa- tients received the amount of crystalloid fluid resuscitation recom- mended by current SCC guidelines. A study period extending beyond the first 3 h of ED admission may have increased the number of patients receiving this obligatory fluid volume of 30 mL/kg TBW.

The necessary information for a sepsis diagnosis is not always avail- able to the treating physician. Our enrolled patients were retrospective- ly identified as septic by the presence of a positive laboratory or other diagnostic test finding of bacterial infection. This method limits the gen- eralizability of our findings, because this laboratory result of a bacterial infection would not be known to the treating physician at the time of the decision to initiate a fluid resuscitation for shock. However, all pa- tients were known to be hypotensive at the time of ED arrival and sepsis was a likely primary consideration for the treating physician consider- ing a perceived high acuity of illness requiring aggressive fluid resusci- tation. Additionally, our data set is not intended to contribute to the diagnostic process. Rather, our investigation sought to identify a non-in- vasive tool that indicated patient improvement or decompensation dur- ing the treatment process. Future studies may compare BPV between septic and non-septic patients in order to evaluate the utility of BPV analysis as a tool that could contribute to identifying multi-organ dys- function, an integral component of a sepsis diagnosis.

Blood pressure variability, specifically among septic patients, is a largely under-studied concept. Pandey et al. found BPV was significantly higher in septic patients with increased Severity of disease measured by Acute Physiology and Chronic Health Evaluation II (APACHE II) scores [15]. In that study, BP was measured at 30-minute intervals during the day and 60-minute intervals at night, over a 24 hour period within the first 48 h of hospital admission. Our study differs in regards to both the frequency of BP measurement (15 min) and the study time period relative to admission (within first 3 h of ED admission). This under- scores the need for data that can be used early in the management of the disease, rather than several days into hospital admission. That study also assessed BPV via standard deviation while our study assesses BPV via ARV. ARV has demonstrated superior prognostic value in the setting of organ damage when compared to the standard deviation method, which does not account for the sequence of values within a data set [24,41].

Just as BPV is an under-studied concept, ARV is a novel statistical tool for measuring BPV. Increasing or decreasing the time interval between BP measurements will result in less or more BP measurements, which consequently can increase or decrease ARV. Our time intervals between BP measurements were determined by expert clinicians. Other experts

may disagree with a 15 minute interval between BP measurements. SSC guidelines recommend frequent BP measurements to guide fluid re- suscitation [6]. How frequent is unclear. A more in-depth understanding of BPV can help determine the optimal frequency of BP measurements necessary to guide clinical intervention for sepsis. Although ARV calculations account for the number of data points included within a series and involve a summation, it is more a measure of variability than an average value as the summation is calculated utilizing the difference between consecutive values within a series, thereby accounting for the change in data points over time. Additionally, since ARV involves the absolute value of the difference between consecutive BP measure- ments, ARV does not account for the directionality of BP changes. A series of BP measurements that increases or decreases repeatedly about a given BP value can be volatile but ARV would indicate low volatility, if ARV did not involve the absolute value of the difference between consecutive measurements. Therefore, ARV appropriately accounts for the extent of BP change. A limitation we encountered with ARV was that ARV calculation is not possible if a patient has one BP measurement during a designated time period (before or after receiving a fluid infusion milestone), because ARV is based upon the difference between consecutive measurements, requiring two BP values. Ultimately, the primary purpose of our study was to investigate whether BPV is associated with clinical improvement or deterioration. Many questions remain on BPV and on what is a high or low BPV. Future studies may establish target values for BPV determined by ARV to help guide sepsis management.

In contrast to the finding of an association between elevated BPV and disease, another study found that mechanically-ventilated septic patients in an intensive care unit-based study exhibited BP oscillation at reduced frequencies, compared to previous reports on healthy, spon- taneously breathing patients [16]. The authors explained that the iden- tified reduction in BP oscillations may be due to reduced sympathetic autonomic activity. By contrast, the majority of patients in our data set were spontaneously breathing. Only nine of the 40 septic patients (23%) were mechanically-ventilated during the study period. Given that positive-pressure ventilation can influence BP, mechanically- ventilated patients are exposed to an additional confounding variable, as compared to spontaneously breathing patients.

The altered vasomotor tone in sepsis pathophysiology is accompa- nied by myocardial depression [42]. Given this change in cardiac function, we also investigated HRV in our septic patient population, finding a positive correlation between HRV and lactate (both initial and subsequent) in patients who received fluid administration N 10 mL/kg TBW. Notably, this patient cohort (fluid administration N 10 mL/kg TBW) represents a slightly earlier point in time than the pa- tient cohort (fluid administration N 20 mL/kg TBW) that demonstrated a positive correlation between BPV and lactate. We also discovered that HRV prior to receiving 20 mL/kg TBW fluid infusion positively correlat- ed with SOFA scores. Similarly, Bohanon et al. found that HRV analysis was more sensitive than conventional vital signs (cardiac output and mean arterial pressure) in confirming a diagnosis of sepsis in neonates [43]. We also found that the correlation between HRV and lactate was less than the correlation between BPV and lactate, suggesting a weaker relationship between HRV and lactate than the relationship between BPV and lactate. This finding is supported by previous studies, which have shown that illness severity has been associated with low HRV, a contrast with the association between illness severity and high BPV [44-45]. Future studies may investigate the utility of analyzing BPV and HRV conjointly to monitor sepsis management.

Conclusion

Despite advances within the medical field, mortality from sepsis remains high, especially among patients with associated systolic hypo- tension and shock [22-23]. Identifying tools that can aid ED clinicians in the early risk-stratification and monitoring of septic patients is likely

to have a high impact on clinical outcomes [3-4]. With the finding of a positive correlation between BPV and both lactate levels and SOFA score, this pilot study introduces BPV analysis as a real-time, non- invasive tool for continuous sepsis monitoring in the ED. Further study is needed to verify the significance of these findings and the reproduc- ibility in a larger and more heterogeneous cohort of patients.

References

1. Xu J, Murphy SL, Kochanek KD, et al. Deaths: final data for 2013. Natl Vital Stat Rep 2016:64(2).
2. Torio C, Moore B. National inpatient hospital costs: the most expensive conditions by Payer, 2013. HCUP statistical brief #204. Rockville, MD: Agency for Healthcare Re- search and Quality; May 2016. http://www.hcup-us.ahrq.gov/reports/statbriefs/ sb204-Most-Expensive-Hospital-Conditions.pdf.
3. Nguyen HB, Rivers EP, Havstad S, et al. Critical Care in the Emergency Department: a physiologic assessment and outcome evaluation. Acad Emerg Med 2007;7(12): 1354-61.
4. Rivers EP, Ahrens T. Improving outcomes for severe sepsis and septic shock: tools for early identification of at-risk patients and treatment protocol implementation. Crit Care Clin 2008;23:S1-S47.
5. Wang HE, Shapiro NI, Angus DC, et al. National estimates of severe sepsis in United states emergency departments. Crit Care Med 2007;25:1928-36.
6. Rhodes A, Evans LE, Alhazzani W, et al. Surviving sepsis campaign: international guidelines for management of severe sepsis and septic shock, 2016. Intensive Care Med Jan 18 2017. https://doi.org/10.1007/s00134-017-4683-6.
7. Singer M, Deutschman CS, Seymour CW, et al. The third international consensus def- inition for sepsis and septic shock (Sepsis-3). JAMA 2016;315(8):801-10.
8. Otto GP, Sossdorf M, Claus RA, et al. The late phase of sepsis is characterized by an increased microbiological burden and Death rate. Crit Care 2011;15:R183.
9. Leone M, Asfar P, Radermacher P, et al. Optimizing mean arterial pressure in septic shock: a critical reappraisal of the literature. Crit Care 2015;19:101.
10. Grassi G, Bombelli M, Brambilla G, et al. Total cardiovascular risk, blood pressure var- iability and adrenergic overdrive in hypertension: evidence, mechanisms and clini- cal implications. Curr Hypertens Rep 2012;14:333-8.
11. Irigoyen MC, Angelis KD, Santos FD, et al. Hypertension, blood pressure variability, and target organ lesion. Curr Hypertens Rep 2016;18:31.
12. Parati G, Ochoa JE, Lombardi C, et al. Blood pressure variability: assessment, predic- tive value and potential as a therapeutic target. Curr Hypertens Rep 2015;17:23.
13. Parati G, Faini A, Valentini M. Blood pressure variability: its measurement and signif- icance in hypertension. J Hypertens 2005;23(1):S19-25.
14. Hocht C. Blood pressure variability: prognostic value and Therapeutic implications. ISRN Hypertens 2013:398485.
15. Pandey NR, Bian Y, Shou S. Significance of blood pressure variability in patients with sepsis. World J Emerg Med 2014;5(1):42-7.
16. Berg RMG, Plovsing RR, Greve AM, et al. Spontaneous blood pressure oscillation in Mechanically ventilated patients with sepsis. Blood Press Monit 2016;21:75-9.
17. Fuller BM, Dellinger RP. Lactate as a hemodynamic marker in the critically ill. Curr Opin Crit Care 2012;18(3):267-72.
18. Casserly B, Phillips GS, Schorr C, et al. Lactate measurement in sepsis-induced tissue hypoperfusion: results from the Surviving Sepsis Campaign Database. Crit Care Med 2015;43(3):567-73.
19. Nguyen BH, Rivers EP, Knoblich BP, et al. Early lactate clearance is associated with improved outcome in severe sepsis and septic shock. Crit Care Med 2004;32(8): 1637-42.
20. Dettmer M, Holthaus CV, Fuller BM. The impact of serial lactate monitoring on emer- gency department resuscitation interventions and clinical outcomes in severe sepsis and septic shock: an observational cohort study. Shock 2015;43(1):55-61.
21. Vincent JL, de Mendonca A, Cantraine F, et al. Use of the SOFA score to assess the in- cidence of organ dysfunction/failure in intensive care units: results of a multicenter,

    prospective study. Working group on “sepsis-related problems” of the European So- ciety of Intensive Care Medicine. Crit Care Med 1998;26(11):1793-800.

    Moore JX, Donnelly JP, Griffin R, et al. Defining sepsis mortality clusters in the United States. Crit Care Med 2016;44(7):138-1387.

22. Ani C, Farshidpanah S, Bellinghausen SA, et al. Variations in organism-specific severe sepsis mortality in the United States: 1999-2008. Crit Care Med 2015;43(1):65-77.
23. Mena L, Pintos S, Queipo NV, et al. A reliable index for the prognostic significance of blood pressure variability. J Hypertens 2005;23(3):505-11.
24. Pandharipande PP, Shintani AK, Hagerman HE, et al. Derivation and validation of SpO2/FiO2 ratio to impute for PaO2/FiO2 ratio in the respiratory component of the Sequential Organ Failure Assessment score. Crit Care Med 2009;37(4): 1317-21.
25. Pandharipande PP, Sanders N, St Jacques P. Calculating SOFA scores when arterial blood gasses are not available: validating SpO2/FiO2 ratios for imputing PaO2/FiO2 ratios in the SOFA Score. Crit Care Med 2006;34(12 suppl):A1.
26. Kolmogorov A, Formin S. Introductory real analysis. 1st edition. New York: Dover Publications; 1975 328.
27. Jones AE, Trezciak S, Kline JA. The Sequential Organ Failure Assessment score for predicting outcome in patients with severe sepsis and evidence of hypoperfusion at the time of emergency department presentation. Crit Care Med 2009;37(5): 1649-54.
28. Gotts JE, Matthay MA. Sepsis: pathophysiology and clinical management. BMJ 2016; 353:i1585.
29. Sharshar T, Gray F, Grandmaison GL, et al. Apoptosis of neurons in cardiovascular au- tonomic centers trigged by inducible nitric oxide synthase after death from septic shock. Lancet 2003;362:1799-805.
30. Toweill DL, Sonnenthal K, Kimberly B, et al. Linear and nonlinear analysis of hemo- dynamic signals during sepsis and septic shock. Crit Care Med 2000;28(6):2051-7.
31. Piepoli M, Garrard CS, Kontoyannnis DA, et al. Autonomic control of the heart and Peripheral vessels in human septic shock. Intensive Care Med 1995;21:112-9.
32. Chillon JM, Baumbach GL. Autoregulation: arterial and intracranial pressure. In: Edvinsson L, Krause DN, editors. Cerebral blood flow and metabolism. Second Edi- tion. Philadelphia: Lippincott, Williams, & Wilkins; 2002. p. 395-412.
33. Carlstrom M, Wilcox CS, Arendshorst WJ. Renal autoregulation in health and disease.

    Physiol Rev 2015;95(2):405-511.

    Schramm P, Klein KU, Falkenber L, et al. Impaired Cerebrovascular autoregulation in patients with severe sepsis and sepsis-associated delirium. Crit Care 2012;16:R181.

34. Blindra J, Pharm P, Chuan A, et al. Is impaired cerebrovascular autoregulation associ- ated with outcome in patients admitted to the ICU with early septic shock? Crit Care Resusc 2016;18(2):95-101.
35. Bagshaw SM, George C, Bellomo R. Early acute kidney injury and sepsis: a multicentre evaluation. Crit Care 2008;12(2):R47.
36. Ergin B, Kapucu A, Demirci-Tansel, et al. The renal microcirculation in sepsis. Nephrol Dial Transplant 2015;30:169-77.
37. Stimac J, Paxton J. The “Golden Hour” of volume resuscitation: pilot data from the Shock Access for Emergent Resuscitation (SAFER) study. Ann Emerg Med 2015; 66(4s):S110.
38. Paxton JH, Courage C, Hunt NR, et al. Initial fluid challenge for hypovolemic septic shock patients: are the new guidelines that much harder? Ann Emerg Med 2014; 64(4s):S71.
39. Pierdomenico S, Nicola MD, Esposito AL, et al. Prognostic value of different indices of blood pressure variability in Hypertensive patients. Am J Hypertens 2009;22(8): 842-7.
40. Fenton KE, Parker MM. Cardiac function and dysfunction in sepsis. Clin Chest Med 2016;37:289-98.
41. Bohanon FJ, Mrazek AA, Shabana MT, et al. Heart rate variability analysis is more sensitive at identifying Neonatal sepsis than conventional vital signs. Am J Surg 2015;210:661-7.
42. Barnaby D, Ferrick K, Kaplan DT, et al. Heart rate variability in emergency depart- ment patients with sepsis. Acad Emerg Med 2002;9(7):661-70.
43. Pontet J, Contreras P, Curbelo A, et al. Heart rate variability as early marker of mul- tiple organ dysfunction syndrome in septic patients. J Crit Care 2003;18(3):156-63.