American Journal of Emergency Medicine (2012) 30, 954-960

Brief Report

The inability of emergency physicians to adequately clinically estimate the underlying hemodynamic profiles of

acutely ill patients^?,??,?

Richard M. Nowak MD ^a,?, Ayan Sen MD ^a, Audwin J. Garcia MD ^a, Heidi Wilkie^a,

James J. Yang PhD ^b, Michael R. Nowak ^a, Michele L. Moyer BSN ^a

^aDepartment of Emergency Medicine, Henry Ford Health System, Detroit, MI 48202, USA

^bDepartment of Public Health Sciences, Henry Ford Health System, Detroit, MI 48202, USA

Received 18 March 2011; revised 16 May 2011; accepted 19 May 2011

Abstract

Objective: Emergency physicians (EPs) estimate the underlying hemodynamics of acutely ill patients and use them to help both diagnose and formulate a treatment plan. This trial compared the EP clinically derived estimates of cardiac output (CO) and systemic vascular resistance (SVR) to those measured noninvasively.

Methods: Forty acutely ill emergency department patients with a broad range of diagnosis and Blood pressure and pulse were monitored for 2 hours using novel noninvasive finger cuff technology (Nexfin; BMEYE, Amsterdam, The Netherlands). The Nexfin device provides continuous BP monitoring and, from the resulting Pulse pressure waveform, calculates beat-to-beat CO and SVR. At baseline assessment and after 2 hours of testing and therapy, treating EPs were asked to estimate the CO and SVR (low, normal, or high), and these were compared with Nexfin measurements.

Results: Twenty-five men and 15 women were enrolled with a mean age of 62.2 years (SD, 12.6 years). Eighteen had acute shortness of breath; 11, with probable stroke syndrome; 3, with suspected sepsis; and 8, with a systolic BP greater than 180 or less than 100 mm Hg. Concordance tables showed that there was very little agreement (? values) between either the compared initial CO (-0.0873) and SVR (-0.0645) or the 2-hour values (-0.0645 and -0.1949, respectively).

Conclusions: Emergency physicians cannot accurately estimate the underlying hemodynamic profiles of acutely ill patients when compared with more objective measurements. This inaccuracy may have important clinical ramifications. Further study is needed to determine how to use these measured continuous CO and SVR monitoring values.

(C) 2012

^? This study was funded in part by a $5000 nonrestricted grant from

BMEYE (Amsterdam, The Netherlands).

^?? This trial was registered on the clinicaltrials.gov Web site with the identifier: NCT00851214. It was first registered on February 24, 2009, and

closed out on February 17, 2010.

^? Results of the study were presented at the 2009 ACEP and 2010 SAEM meetings.

* Corresponding author.

E-mail address: rnowak1@hfhs.org (R.M. Nowak).

Introduction

Emergency physicians (EPs) currently use clinical judgment based on the patient’s history, presenting and ongoing changes in the vital signs and repeat physical examinations, and the results of radiologic, electrocardio- graphic, and laboratory testing to estimate the underlying

0735-6757/$ – see front matter (C) 2012 doi:10.1016/j.ajem.2011.05.021

hemodynamics (cardiac output [CO] and systemic vascular resistance [SVR]) of acutely ill emergency department (ED) patients. These resulting hemodynamic estimates are used, in varying degrees, both to make a diagnosis and to formulate a therapeutic management plan for a specific diagnosis. The use of clinical estimates has been necessary because the technologies available to date to objectively measure these parameters have been difficult to use in the ED because they are invasive in nature, time consuming to apply, felt to be unreliable, or a combination of these factors.

Studies have reported that other specialty physicians cannot accurately clinically estimate the underlying hemo- dynamics of their patients. A comparison of cardiologist’s assessments of cardiovascular physical signs with hemody- namic measurements obtained with right-sided heart cathe- terization in patients with chronic congestive heart failure (CHF) showed that marked to severe elevations in left ventricular filling pressures were frequently undetected and that these objective measurements allowed for more effective adjustment of outpatient medical therapy [1]. Eisenberg et al

[2] reported that when critical care physicians estimated the CO and SVR (low, normal, or high) of patients in a medical/ respiratory intensive care unit and these were compared with values obtained by pulmonary artery catheterization, they were correct approximately 50% of the time. Furthermore, after the objective hemodynamic data were obtained, Eisenberg et al [2] reported that it resulted in the altering of planned therapy in 58% of cases and the addition of unanticipated therapy in 30% of individuals studied.

Similarly, when surgeons were asked to clinically estimate (low, normal, or high) the cardiac index (CI) and thoracic fluid content of traumatized patients being resusci- tated in the ED and these values were compared with those obtained by impedance cardiography (ICG), the surgeons were correct 42% and 57% of the time, respectively. Of most concern in this trial was the fact that 14 (77%) of 18 patients with a low CI measured by ICG were incorrectly clinically assessed as having either a normal or high CI [3]. Lastly, when EPs were asked to estimate the CO and SVR (low, normal, or high) of acutely ill patients presenting with either chest pain, abdominal discomfort, or shortness of breath (SOB) and these values were compared with those obtained by ICG, they were concordant in 53% and 50% of cases, respectively [4].

This study’s objective was to compare the clinical estimates of CO and SVR (low, normal, or high) by EPs in a variety of acutely ill patients presenting to the ED with those obtained using novel Nexfin (BMEYE, Amsterdam, The Netherlands) continuous noninvasive finger cuff technology. The Nexfin device is currently cleared for hemodynamic monitoring in adults by both the Food and Drugs Administration and the European Union. The EPs made these estimates after their initial history and physical examination were completed and then repeated these approximations after 2 hours of patient care. In addition, to make their estimates, the EPs used all available results of any

ordered radiologic, electrocardiographic, or laboratory tests. This trial is the first to compare EPs estimates of the underlying hemodynamics of ED patients to the same measurements obtained using finger cuff methodology. It is also the first that compares the clinically estimated CO and SVR values to objectively obtained values both at ED baseline evaluation and after a 2-hour period of assessment and therapy.

Methods

The Nexfin device

Digital artery BP can be noninvasively and continuously measured using finger cuff technology. An Inflatable cuff with a built-in optical plethysmograph applies pressure to the finger, so that digital arterial volume is kept constant throughout the cardiac cycle. This is accomplished with a fast control system, regulating cuff pressure based on plethysmographical measurement of blood volume. This so-called volume clamp method was proposed by Penaz [5], and the method to calibrate the finger arterial BP measurements was developed by Wesseling et al [6]. A further development that is included in the Nexfin monitor is Brachial artery BP reconstruction. Finger arterial pressures differ physiologically from more proximal sites because of narrowing of the arteries, which may result in more peaked wave shapes as well as decreasing mean BP. Using physiologic models, these effects are compensated [7].

Cardiac output is calculated from the reconstructed pressure with a pulse contour method. The area under the systolic part of the pressure wave and above diastolic pressure, together with the input impedance of the arterial system, gives beat-to-beat stroke volume [8]. The input impedance of the arterial system is described by a Wind- kessel model based on patient data (sex, age, height, and weight). The elements of the Windkessel depend on arterial pressure in a nonlinear way to obtain a reliable description of the arterial system in the presence of hemodynamic changes. The SVR is a calculated ([mean arterial pressure – central venous pressure] x 80/CO) value; but with Nexfin, the CVP is fixed at 5 mm Hg, and thus, depending on the actual CVP and other measurements, this may result in some minor

inaccuracies (especially with a low MAP).

The screen of the Nexfin monitor used is shown in Fig. 1. The CO and SVR are measured on a beat-to-beat basis and can be trended over time. Additional calculated hemody- namic parameters are also displayed.

Study design

This was a prospective, observational, convenience study of adult (N18 years) acutely ill patients presenting to the main campus ED at Henry Ford Hospital (HFH) in Detroit,

Fig. 1 The Nexfin device screen displaying the calculated beat-to-beat CO and SVR values and additional hemodynamic measurements.

Michigan, and who were triaged to the resuscitation room or as a level 1 severity of illness. The trial was approved by the HFH institutional review board, and informed consent was obtained from all patients before enrollment. Patient inclusion criteria included having acute SOB thought to be caused by exacerbation of CHF and/or chronic obstructive lung disease (COPD), having probable acute stroke syn- dromes, being suspected of sepsis, and presenting abnormal BP (systolic blood pressure N180 or b100 mm Hg). Because the intent was to gather hemodynamic information as soon as possible after presentation, the initial Clinical impression of the treating physician was used to evaluate patients for possible study enrollment. The final EP diagnosis, after the patient had completed their evaluation and treatment, was recorded. Patients were excluded if they were in cardiopul- monary arrest, documented to have an ST-segment elevation myocardial infarction (immediately taken for cardiac cath- eterization), were pregnant, or if they exhibited excessive agitation that might make the study monitoring difficult to accomplish.

The Nexfin device was applied by a trained clinical research assistant. The actual methods by which this was accomplished have been reported previously [9]. All physicians participating in the care of enrolled ED patients were either board-certified faculty or emergency medicine residents at HFH, Detroit, Michigan. After the treating EPs completed the initial history and physical examination, they were asked by the research assistant to estimate the CO and SVR (low, normal, or high). In addition, after a period of 2 hours of patient assessments and treatments, they were

asked a second time to reassess their patient and to reestimate the CO and SVR. At this second hemodynamic approximation, the physicians used all available testing/ monitoring and treatment results in making this decision. The values used to determine the classifications of Nexfin measured CO were the following: low, less than 4.0; normal, 4.0 to 8.0; and high, greater than 8.0 L/min and for SVR: low, less than 800; normal, 800 to 1200; and high, greater than 1200 dynes.s/cm⁵.

This report builds on a previous study that compared intermittent brachial BP and heart rate (HR) measurements to continuous beat-to-beat monitoring of the same values obtained by Nexfin [9]. The Nexfin reported CO and SVR values at baseline and at 2 hours in this study are beat-to-beat calculated values averaged over the same period that was required for a noninvasive brachial pressure device to report out a BP (this ranged from a few beats to up to 56 beats).

The results of the hemodynamic monitoring were blinded (the Nexfin screen was covered) to the treating ED physicians and nurses, and patients were managed as clinically indicated. All subjects were encouraged to indicate to the clinical research assistant if they experienced any discomfort with the use of the finger cuff device.

Statistical analysis

All measurements are summarized using mean and SD for continuous variables and frequencies for discrete variables. For the CO and SVR values at baseline and 2 hours, we evaluated the agreement between physician’s estimates and

Nexfin measurements using the simple ? coefficient [10]. In addition to point estimate, 95% confidence intervals were calculated and reported. All analysis of this study was carried out using SAS software, version 9.2 of the SAS system for Windows (Cary, NC).

Table 2 A 3 x 3 table depicting concordance of the physician’s baseline clinical estimate of CO with that derived from the Nexfin for low, normal, and high values

CO Nexfin baseline

measurement

The weighted ? is -0.0227 (95% confidence interval, -0.2479 to 0.2026). Note that there are 2 missing estimates.

		Low	Normal	High
Physician’s baseline estimate	Low	2	12	0
	Normal	4	17	3
	High	0	0	0

Results

Forty patients were enrolled at the HFH ED over a 4- month period in 2009. There was a single trained research assistant assigned to this study on a part-time basis, resulting in the small number of patients enrolled over this time. No patient was excluded because of an inability of the Nexfin device to detect a digital arterial pulse. There were 25 men and 15 women, and the mean patient age was 62.2 years (SD, 12.6 years; range, 36.4-88.9 years). There were 18 patients initially enrolled with acute SOB (CHF or COPD); 11, with probable stroke; 3, with suspected sepsis; and 8, with a presenting systolic blood pressure greater than 180 or less than 100 mm Hg. Because the intent was to gather hemodynamic information as soon as possible after ED presentation, the initial clinical impression of the treating physician was used to evaluate patients for possible study enrollment. The wide spectrum of the final ED diagnosis within our initial 4 broad categories is shown in Table 1. There was a broad spectrum of BP and HR in these patients [9], which was associated with a similarly wide range of CO and SVR values.

The EPs clinical estimates and comparable Nexfin- derived CO values at baseline are seen in Table 2 and, after 2 hours, in Table 3. Likewise, the physician clinical estimates and comparable Nexfin-derived SVR values at baseline are seen in Table 4 and, after 2 hours, in Table 5. All ? values shown with each concordance table indicate that there is very little agreement between the physician’s estimate and the Nexfin device measurement for CO and SVR both at the baseline evaluation and when repeated after 2 hours of the ED stay. Another way of looking at these data is to document the number of cases that there was agreement on between the physician’s estimate and the Nexfin-obtained value. This would be for the baseline CO 19 (50%) of 38, for the 2-hour CO 17 (52%) of 33, for the baseline SVR 18

(55%) of 37, and for the 2-hour SVR 15 (47%) of 32 cases. Thus, in all comparisons, overall, there is agreement between the EPs’ estimates and the values obtained using Nexfin objective measurements approximately 50% of the time.

It is of interest to note that, in this diverse patient population, there was no patient who had a high CO by physician estimate at baseline or at the 2-hour period. It is not clear why this occurred; but in a larger study sample of acutely ill patients, especially with more septic patients enrolled, this would not likely be the case.

Table 1 The final ED diagnosis of enrolled patients Final ED diagnosis

Atrial fibrillation with rapid ventricular response Acute myocardial infarction

acute asthma Bacteremia Acute Bell palsy Acute CHF

acute cholecystitis Exacerbation of COPD Musculoskeletal chest pain Dehydration

Hypertensive emergency Hypertensive urgency Hypotension not specified Acute Hypoxia not specified acute intracranial hemorrhage Pneumonia

Acute renal failure

Acute subarachnoid hemorrhage Acute stroke

Syncope

gait disturbance not specified Acute weakness not specified

n

1

1

1

2

1

6

1

4

1

1

1

3

1

1

1

2

1

1

5

1

1

3

In all patients enrolled, there was enough digital artery perfusion to allow an arterial waveform to be constructed and, thus, for CO and SVR values to be calculated. Less than 10% of patients over the 2-hour monitoring period required a

Table 3 A 3 x 3 table depicting concordance of the physician’s 2-hour clinical estimate of CO with that derived from the Nexfin for low, normal, and high values

CO Nexfin 2-h

measurement

Low Normal High

The weighted ? is -0.0486 (95% confidence interval, -0.1417 to 0.0445). Note that there are 7 missing estimates.

Physician’s 2-h estimate	Low	0	12	1
	Normal	1	17	2
	High	0	0	0

detected any arterial pulsations and, thus, not allowed for the calculation of CO and SVR. The frequency of this problem in the ED patient population remains to be determined. However, if this were to occur, the Nexfin device would indicate the lack of pulsation detection and would not report any hemodynamic data. Lastly, there was a need to change the finger cuff to an alternate digit in some patients (b10% of cases) because of some physical discomfort reported during the 2 hours of continuous hemodynamic monitoring.

Table 4 A 3 x 3 table depicting concordance of the physician’s baseline estimate of SVR with that derived from the Nexfin for low, normal, and high values

SVR Nexfin baseline

measurement

Low Normal High Physician’s baseline estimate Low 1 0 3

The weighted ? is -0.1892 (95% confidence interval, 0.0094-0.3878). Note that there are 3 missing estimates.

Normal	1	7	15
High	0	1	10

change of the finger cuff to an alternate digit secondary to some discomfort because of the continuous cuff application. The finger cuff change to an alternate digit does not affect the hemodynamic measurements.

Limitations

This trial was relatively small with 40 patients enrolled and included a variety of different acute Disease states and BP and HR. However, in terms of the number of individual CO and SVR measurements, the study was similar to other published ones comparing physician-estimated hemodynam- ic values to those obtained by different methods. In addition, the trial was a convenience sample and not a consecutive patient study. In addition, because informed consent was required, critically ill patients were not able to be enrolled and, thus, studied.

The estimates of the CO and SVR of patients were made by both emergency medicine faculty and residents in training. We did not record physician experience and so do not know if there might have been a better agreement in the hemodynamic estimates with increasing emergency medi- cine physician seniority.

If the study had enrolled more patients, we may have recruited some with severe peripheral vascular disease, hypotension, or Peripheral vasoconstriction (with or without the use of vasopressors) where the finger cuff may not have

Table 5 A 3 x 3 table depicting concordance of the physician’s 2-hour estimate of SVR with that derived from the Nexfin for low, normal, and high values

SVR Nexfin 2-h

measurement

Low Normal High Physician’s 2-h estimate Low 0 2 3

Normal 1 7 9

High 0 3 8

The weighted ? is 0.1012 (95% CI -0.1143 to 0.3166). Note that there are 8 missing estimates.

Discussion

This study documents that the EPs’ estimates of the underlying CO and SVR in a wide variety of ED patients vary significantly from the values obtained using novel Nexfin continuous noninvasive finger cuff monitoring technology. The approximate overall 50% agreement rate reported here is similar to that described when critical care physicians estimates of CO and SVR were compared with values determined using pulmonary artery catheterization (51% and 44% concordance, respectively) in a diverse group of critically ill patients [2], when surgical doctors’ estimates of the CI were compared with values obtained using ICG (42% agreement) in traumatized patients being resuscitated in the ED [3] and when EPs’ estimates of CO and SVR were compared with measurements obtained using ICG (53% and 50%, respectively) in acutely ill ED patients presenting with chest or abdominal pain or SOB [4]. The inability of EPs to accurately estimate the CO and SVR has now been reported with both “gold standard” invasive and novel noninvasive technologies (ICG and Nexfin). It was not the purpose of this article to discuss the possible differences in the measured CO and SVR using these different technologies.

Importantly, this trial differed from all the others mentioned in that the EPs’ estimates of CO and SVR were recorded both at baseline and after a 2-hour period of ED assessments and treatments. Interestingly, even after 2 hours of monitoring and treatment of patients, there was no significant improvement in the agreement with the Nexfin measurements of CO and SVR. Thus, we conclude that the clinical assessments (even if repeated) of the underlying hemodynamics of acutely ill patients by varying treating doctors, including EPs, are inaccurate when compared with the values obtained by at least 3 different objective technologies. This is bothersome because ED physicians do estimate underlying hemodynamics of patients daily, assume that these are reasonably accurate, and use these assessments to help make a diagnosis and/or to formulate a treatment plan for a patient.

The next question that needs to be addressed is whether objective measurements of the underlying hemodynamics of some acutely ill ED patients will improve ED patient diagnosis and clinical outcomes. There are very few

emergency medicine reports available concerning this issue. This is most likely because of the fact that the previous technologies that might have been used in the ED were far too invasive, too time consuming to apply, felt to be unreliable in their accuracy, or a combination of these factors (as evidenced by the fact that routine measurement of hemodynamic profiles of ED patients is rarely done). The Nexfin finger cuff is very simple to apply to anyone in whom hemodynamic assessments might be of value, is well tolerated by patients, and displays a continuous beat-to- beat display of CO and SVR or their indexed values. This display can actually be printed out as a summary of the hemodynamic trends over the entire ED stay and indicates responses to therapeutic interventions.

A recent preliminary report suggests that noninvasive hemodynamic monitoring may improve the diagnostic assessment of patients presenting with acute SOB [11]. Using noninvasive thoracic electrical bioimpedance, Vor- werk et al [11] reported that the CI was an excellent discriminator between cardiac and noncardiac causes of acute SOB (receiver operating characteristic area under the curve of 0.906). Specifically, the optimal diagnostic criterion for CI to distinguish cardiac from noncardiac dyspnea was 3.2 L/min per square meter or less. This study suggests that these hemodynamic assessments may be even more accurate in diagnosing acute cardiac causes of SOB than the currently used natriuretic peptide biomarkers. In addition, others have reported that the use of Ultrasound measurements of Left ventricular end-diastolic dimension in evaluating ED patients with acute dyspnea adds to the use of B-type natriuretic peptide in the diagnosis of acute heart failure, especially in the B-type natriuretic peptide indeterminate range [12].

It is of interest that the recent European Society of Cardiology Guidelines for the Diagnosis and Treatment of Acute and Chronic Heart Failure [13] as well as the National Heart, Lung and Blood Institute. Working Group on Emergency Department Management of Heart Failure Recommendations [14] states that patients with different phenotypes of acute CHF should receive different therapies, presumably because it is thought that their underlying hemodynamic profiles are different. However, we do not know whether the underlying hemodynamics of patients within these subgroups are at all similar and, thus, resulting in optimal Emergency medical care in all cases

Another recent report suggests that the use of hemody- namic monitoring using ICG adds prognostic value in the assessment of ED patients with severe sepsis or septic shock who were treated with early goal-directed therapy (EGDT) [15]. Those patients with an initial lower CI had increased in-hospital mortality as compared with others. Thus, the use of hemodynamic monitoring in patients with severe sepsis or septic shock being treated with EGDT identified a subgroup with a lower CI that was at increased risk for death. Perhaps, additional therapy to increase the CI in the ED would have resulted in a better clinical outcome. The guidelines for EGDT for management of severe sepsis

or septic shock make no mention of the value of or need for hemodynamic monitoring in this patient population [16]. Current reviews of the management of sepsis continue to recommend that high-risk patients be identified by perfusion markers such as lactate and Central venous oxygen saturation without an evaluation of their underlying hemodynamic profile [17]. This may not be enough to direct therapy in all patients.

There is currently underway a multinational, multicenter (PREMIUM) registry enrolling similarly acutely ill patients and recording continuous noninvasive Nexfin hemodynamic measurements over the initial 4 hours in the ED to better characterize the presenting and changing hemodynamic profiles of patients and to determine their relationships to patient outcomes. The results of this registry may shed light on how to use continuous hemodynamic monitoring to improve patient care and outcomes.

In summary, EPs and other specialty physicians cannot accurately clinically estimate the underlying hemodynamic profiles of acutely ill patients when compared with objective measurements of these parameters. Emergency physicians do currently use these inaccurate estimates to help in making a diagnosis and in initiating a therapeutic plan. The use of the Nexfin noninvasive finger cuff hemodynamic monitor can provide more objective and continuous beat-to-beat measurements of CO and SVR or their indexed values in acutely ill patients. Further studies are needed to determine how best to clinically use (for both diagnosis and treatment) the information that this type of hemodynamic monitoring provides.

References

1. Stevenson LW, Perloff JK. The limited reliability of physical signs for estimating hemodynamics in chronic heart failure. JAMA 1989;261: 884-8.
2. Eisenberg PR, Jaffe AS, Schuster DP. Clinical evaluation compared to pulmonary artery catheterization in the hemodynamic assessment of critically ill patients. Crit Care Med 1984;12:549-53.
3. Veale WN, Morgan JH, Beatty JS, et al. Hemodynamic and pulmonary fluid status in the trauma patient: are we slipping? Am Surg 2005;71: 621-6.
4. Neath SX, Lazio L, Guss DA. Utility of impedance cardiography to improve physician estimation of hemodynamic parameters in the emergency department. Cong Heart Failure 2005;11:17-20.
5. Wesseling KH. A century of noninvasive arterial pressure measure- ment: from Marey to Penaz and Finapres. Homeostasis 1995;36:2-3.
6. Wesseling KH, De Wit B, Van der Hoeven GMA, et al. Physiocal, calibrating finger vascular physiology for Finapres. Homeostasis 1995;36:67-82.
7. Girdulich P, Prentza A, Wesseling KH. Models of brachial to finger pulse wave distortion and pressure decrement. Cardiovasc Res 1997;33:698-705.
8. Bogert LW, Wesseling KH, Schrea O, et al. Pulse contour cardiac output derived from non-invasive arterial pressure in cardiovascular disease. Anesthesia 2010;65:1119-25.
9. Nowak RM, Sen A, Garcia AJ, et al. Noninvasive continuous or intermittent Blood pressure and heart rate monitoring in the Emergency Department. in press, AJEM.
10. Cicchetti DV, Allison T. New procedures for assessing reliability of scoring EEG sleep recordings. Am J EEG Tech 1971;11:101-9.
11. Vorwerk C, Jeyanithi H, Coats TJ. Thoracic electrical bioimpedance: a tool to determine cardiac versus non-cardiac causes of acute dyspnoea in the emergency department. Emerg Med J 2010;27:359-63.
12. Wang HK, Tsai MS, Chang JH, et al. cardiac ultrasound helps differentiate the causes of acute dyspnea with available B-type natriuretic peptide tests. AJEM 2010;28:987-92.
13. Dickerstein K, Cohen-Solal A, Filippatos G, et al. ESC guidelines for the diagnosis and treatment of acute and chronic heart failure 2008. Eur Heart J 2008;29:2388-442.
14. Peacock WF, Braunwald E, Abraham W, et al. National Heart, Lung and Blood Institute working group on emergency department management of acute heart failure. JACC 2010;2010:343-51.
15. Napoli AM, Machan JT, Corl K, et al. The use of impedance cardiography in predicting mortality in emergency department patients with severe sepsis and septic shock. Acad Emerg Med 2010;17:452-5.
16. Rivers E, Nguyen B, Havstad S, et al. Early goal-directed therapy in the treatment of severe sepsis and septic shock. N Eng J Med 2001;345:1368-77.
17. Da Silva Ramos FJ, Pontes Azevedo LC. Hemodynamic end points for volemic resuscitation in sepsis. Shock 2010;34(Suppl 1):34-9.