ECC EDUCATION

Optimizing Oxygen Delivery in the Critically Ill: Assessment of Volume Responsiveness in the Septic Patient

Benjamin de Witt, MD, Raj Joshi, MD, Harvey Meislin, MD, Jarrod M. Mosier, MD
The Journal of Emergency Medicine; Vol 47, Issue 5. November 2014, pg 608-615

- - - - -

Introduction
Critically ill patients with circulatory failure are commonly encountered in the emergency department (ED) setting, and initial resuscitation often involves an aggressive approach to volume expansion. The ultimate goal of volume expansion is to improve cardiac output and ultimately, oxygen delivery. Whereas under-resuscitation of the hemodynamically compromised patient can result in further end-organ dysfunction, over-resuscitation can lead to volume overload and prolong the need for mechanical ventilation, intensive care unit length of stay, and increase mortality. Assessment of volume status, or more appropriately, volume responsiveness (i.e., whether a patient's cardiac index will be responsive to increased circulatory volume) is a skill of paramount importance to the emergency physician, so that resuscitation may be performed in a rational manner. The means of assessing volume responsiveness have been the subject of great controversy and active research. This review will describe the currently available methods of volume assessment, with a focus on the septic patient in circulatory failure.

Discussion

Cardiac Filling Pressures and the Single-Pump Model of the Circulation
Cardiac filling pressures (central venous pressure [CVP] and pulmonary artery occlusion pressure [PAOP]) have been used as a guide for optimizing preload and fluid resuscitation since the 1950s, as first described by Hughes and Magovern. Cardiac filling pressures are thought to obey the Starling principle, which states that increased stretch (preload) of a cardiac myocyte will increase the contractility until you reach a plateau (Figure 1). When applied to the cardiovascular system of a patient in shock rather than a single myocyte or cardiac chamber, the assumption is that changes in right atrial pressure (preload) lead to changes in cardiac output (contractility), and has been shown to be true in healthy male volunteers as central venous pressure changes linearly with either hemorrhage or transfusion for the first several minutes.


Figure 1. The “Starling Curve” demonstrating changes in contractility per change in preload. 

The “Starling Curve” states that as you increase the preload, or stretch, of a muscle (i.e., left ventricle) the output will increase. Once the preload limit is reached, further increases in preload lead to only minimal increases in output. Adapted from Michard F, Teboul JL. Using heart-lung interactions to assess fluid responsiveness during mechanical ventilation. Crit Care 2000;4:282–9, Figure 1.

This is consistent with a single-pump model of the circulation, which ignores the right ventricle and pulmonary circulatory effects on left ventricular output, and assumes that restoration of central venous pressure leads to restoration of circulating volume. Previous researchers subscribing to this model report only three possible explanations for shock: loss of vascular tone, decreased contractility, and volume depletion 8 and 9. Thus CVP has gained popularity as the endpoint of choice for volume resuscitation and is incorporated into management algorithms and guidelines for the care of the septic shock patient.

Unfortunately, cardiac filling pressures have been unable to differentiate volume responders from nonresponders with an area under the receiver operating characteristic (ROC) curve for CVP 0.58 and PAOP 0.63. With an area under the ROC curve of 0.58, CVP is only slightly better than chance at predicting volume responders from nonresponders, with chance being 0.50 and a perfect prediction 1.0. Additionally, achieving target cardiac filling pressures shows poor correlation to improvement in either left ventricular performance (end-diastolic volume [EDV], stroke volume [SV], or cardiac index [CI]) or circulating blood volume. Lastly, even CVP at the extremes of range (low or high) fails to differentiate responders from nonresponders. In a comprehensive literature review up to 2006, and repeated in 2012, Marik et al. conclude that the measurement of CVP, whether to determine circulating blood volume or as a response to volume challenge, is not useful, with very poor test characteristics (ROC of 0.56).

These data suggest that cardiac filling pressures are not effective parameters for assessing whether a hemodynamically compromised patient will respond to volume infusion. Rather than a single pump with CVP affected by only three variables, the right ventricle and pulmonary circulation in fact do play an important role. Thus, right ventricular compliance, the presence of tricuspid regurgitation or stenosis, and pulmonary hypertension all influence central venous pressure. Additionally, left ventricular compliance, intrathoracic pressure, chest wall compliance, and vascular tone also contribute to central venous pressure, exclusive of intravascular volume. Given these confounders, cardiac filling pressures cannot reflect the actual right and left end diastolic volumes exclusively, and thus will not be able to predict volume responsiveness clinically.

Cardiopulmonary Interactions and the Dual-Pump Model of the Circulation
A dual-pump model that accounts for right ventricular, pulmonary vascular, and intrathoracic influence on cardiac performance is a more accurate description of the forces that determine volume responsiveness. The right ventricle (RV) is a low-pressure, highly compliant, flow-based chamber, as opposed to the left ventricle (LV), which is a high-pressure, low-compliance chamber. As a result, the RV is far more sensitive to changes in afterload than the LV, which can maintain stroke volume in the face of large increases in systemic vascular resistance. RV stroke volume, however, is sensitive to variation in venous return, and RV afterload induced by intrathoracic pressure changes during respiration. These fluctuations in RV stroke volume alter LV preload after transit through the pulmonary circulation and ultimately affect LV stroke volume (Figure 2). These fluctuations are termed cardiopulmonary interactions and are the physiologic basis behind all nonfilling pressure methods of volume assessment.


Figure 2. Physiologic effects of cardiopulmonary interactions on stroke volume. RV = right ventricle; LV = left ventricle.

Figure 2 describes cardiopulmonary interactions on stroke volume in both spontaneously breathing and mechanically ventilated patients. In the spontaneously breathing patient, decreased intrathoracic pressure with inspiration will cause an increase in venous return to the RV, leading to a larger stroke volume if the patient is on the ascending portion of the Starling curve. However, in mechanically ventilated patients, the physiology is reversed and an increase in intrathoracic pressure during a ventilator-delivered breath decreases RV preload and increases RV afterload. These transient RV stroke volume fluctuations result in LV stroke volume changes after traveling through the pulmonary circulation and can be monitored both invasively and noninvasively.

Cardiopulmonary Interactions: Invasive Monitoring
Invasively, cardiopulmonary interactions are monitored via arterial line waveform analysis in the form of stroke volume variation (SVV), pulse pressure variation (PPV), and systolic pressure variation (SPV) (Figure 3):

%SVV=(SVmax−SVmin)/[(SVmax+SVmin)/2]×100%,
%PPV=(PPmax−PPmin)/[(PPmax+PPmin)/2]×100%,
SPV=SPmax−SPmin


Figure 3. Stroke volume variation, pulse pressure variation, and systolic pressure variation. Derived from an arterial line tracing.

Pressure variation analysis has been shown to predict volume responsiveness with excellent reliability. When using a pressure variability cutoff of 11–13% over a single respiratory cycle, studies show sensitivity and specificity > 90% and ROC of 0.91–0.99. These data have been shown in mechanically ventilated patients with septic shock as well as multiple other patient groups.

There are significant limitations to using pressure variability analysis. The primary limitation is the need for invasive arterial access. Additionally, the patient must also be in a sinus rhythm. Using advanced algorithms, SVV has been shown to be effective in predicting volume responsiveness in the setting of frequent premature ventricular contractions. An important caveat about pressure variation analyses is that they have only been studied in mechanically ventilated patients. Although stroke volume variation measurement may be feasible in the ED, it requires a central venous catheter with a special monitor to calculate SVV in addition to an arterial line. These extra procedures and special equipment often make such measurements impractical in the ED. However, as described below, SVV can be evaluated noninvasively with bedside echocardiography.

Cardiopulmonary Interactions: Non-invasive Monitoring

Ultrasound
Ultrasonographic assessment of volume responsiveness is becoming increasingly popular in the ED, as invasive monitoring requires time-consuming and often expensive invasive access. The inferior vena cava (IVC) is a compliant, low-pressure vessel with physiologic variations in diameter secondary to cardiopulmonary interactions. Measurement of these changes provides a rapid, although somewhat limited, means of assessing volume responsiveness.

The IVC can be identified easily and quickly, and measured by placing the ultrasound probe in the subxiphoid position, and obtaining a longitudinal view as it courses through the liver. The IVC diameter is measured using M-mode during inspiration and expiration, and is most commonly measured just upstream from the confluence of the hepatic vein and IVC.

In mechanically ventilated patients where intrathoracic pressure changes are reversed as described above, IVC ultrasonography measuring the change in diameter with inspiration has also been shown in small studies to be useful in the setting of mechanically ventilated patients in septic shock. IVC distensibility index (dIVC) is defined as the IVC diameter at end inspiration (Dmax) – the IVC diameter at end inspiration (Dmin) over Dmin [(Dmax – Dmin) /Dmin]. Barbier et al. found that dIVC ≥ 18% is useful for predicting volume responsiveness in mechanically ventilated patients in septic shock. Feissel et al. found similar results when using a dIVC threshold of ≥12% to predict an increase in cardiac index of 15% after a fluid bolus. Contrarily, a recent study by Corl et al. found IVC measurement unable to predict fluid responsiveness in their small sample of ED patients.

Although IVC ultrasound does have data supporting it as outlined above, it does have limitations in several disease states. These include cirrhotic liver disease and situations involving high right-sided pressure, such as pulmonary embolism or acute respiratory distress syndrome. The interpretation of respiratory variation in IVC measurements will also be limited in intubated patients requiring high positive end-expiratory pressure or airway pressure release ventilation, and in patients with a high work of breathing generating large changes in intrathoracic pressure, such as diabetic ketoacidosis.

Although IVC assessment can be limited, cardiopulmonary interactions can also be assessed by ultrasound by evaluating left ventricular physiologic changes to intrathoracic pressure variation, and potentially by measuring large-vessel (carotid, brachial artery) flow variation. On an apical five-chamber view showing the left ventricular outflow tract (LVOT), pulsed-wave Doppler signal provides a noninvasive measurement of stroke volume variation. The waveform is traced to measure the area under the curve (velocity-time integral [VTI]) of each heartbeat at peak inspiration and expiration to determine the percent change. To measure a cardiac output, one must also measure the LVOT diameter in the parasternal long axis view. Applying the equation (π [LVOT diameter/2]2 × VTI) will provide the stroke volume, and multiplying by the heart rate will provide the cardiac output. LVOT VTI variation measurements have been shown to predict volume responsiveness with excellent performance (ROC 0.95) while using a variation threshold of 9% 61 and 62. Simplified measurements of LVOT peak velocity variation (ΔVpeak), the Doppler corollary of systolic pressure variation, have performed very well in predicting volume responsiveness in mechanically ventilated patients with septic shock. Similar analyses on the brachial and carotid arteries show promise, although data are limited and should not yet be used clinically.

Passive leg raise
Whereas the above dynamic measurements attempt to predict response to a volume infusion, the passive leg raise maneuver (PLR) provides a rapid clinical examination method of predicting a patient’s response. When sitting a patient up at 45°, then lying them flat and raising their legs to 45°, the PLR maneuver simulates an approximate 250-mL bolus. Although not without its limitations, the PLR maneuver can predict volume responsiveness when used alone, and when combined with other previously described maneuvers (IVC ultrasound, LVOT VTI, etc.). In situations of low pulmonary compliance (e.g., acute respiratory distress syndrome) that limit the ability to use variation measurements, PLR has been shown in one small study to outperform PPV when combined with an end-expiratory breath-hold on the ventilator. PLR has also been combined with end-tidal CO2 to better discriminate responders at the tissue level from nonresponders.

Summary and Practical Considerations
In the ED, no single method of volume assessment is ideal, due to practical considerations including variability in monitoring capabilities, skills required for and availability of equipment, and patient characteristics such as high positive end-expiratory pressure required for poor lung compliance or dysrhythmias that limit interpretation. Each method has strengths and weaknesses to consider. Arterial waveform analyses require invasive arterial access, and, in the case of SVV, special monitoring capabilities outside the abilities of many EDs.

Ultrasound assessment of cardiopulmonary interactions are limited by patient characteristics such as high right-sided pressures, liver disease, and aggressive ventilator settings in the case of IVC assessment, advanced skills required and body habitus limitations for LVOT VTI, and peak velocity variations and limited data available for carotid and brachial indices.

Passive leg raise can be useful when used alone, and can enhance the ability of the previously described methods to discriminate responders from nonresponders. It is useful in those patients with dysrhythmias, or those who are spontaneously breathing, two significant limitations of other dynamic analyses. A caution to PLR when used alone is to guarantee that the rise in blood pressure is from an increase in intravascular volume rather than an increase in systemic vascular resistance due to a pain or agitation response.

One deficiency in many of these studies is their small number of studied patients. The majority of these studies have fewer than 50 enrolled subjects, and many were performed outside of the ED. Although studies for each modality are relatively consistent in terms of their results, their applicability to the septic shock patient in the ED needs cautious application, as none have been studied with other sepsis care bundles or outcomes-based trials. Nevertheless, dynamic assessments of cardiopulmonary interactions to predict volume responsiveness have consistently outperformed cardiac filling pressures. In summary, we recommend the following methods of volume assessment in the septic shock patient in the ED:

  • If an ultrasound capable of cardiac imaging is available, and the practitioner has the required skill at advanced imaging, LVOT VTI provides the best performance in both mechanically ventilated and spontaneously breathing patients, and can be augmented with passive leg raise. Although they show promise, until carotid and brachial velocities are studied further, they should not be used clinically as a substitute for LVOT VTI.
  • If advanced cardiac imaging is not possible, systolic pressure variation on arterial line waveform should be measured, as it does not require specialized software interpretation, and can be assessed quickly at the bedside. One limitation however, is that this has not been studied in spontaneously breathing patients.
  • Although IVC ultrasound imaging has its limitations, including spontaneously breathing patients, it can be performed by novice ultrasonographers and does have data to support its use.
  • The passive leg raise maneuver may be helpful if ultrasound or systolic pressure variation analyses are not feasible. PLR can be useful when used with the above modalities to augment discriminating responders from nonresponders.
  • Although cardiac filling pressures have been strongly shown not to correlate with predicting volume responsiveness, CVP remains a part of the existing management bundles for septic shock that seem to have improved mortality. CVP should be used in conjunction with interpretation of the patient's underlying physiology and in combination with passive leg raise or gentle boluses rather than a static target. For example, a patient with profound acute respiratory distress syndrome may have a CVP of 12 due to hypoxic vasoconstriction causing high right-sided pressures, but still prove to be volume responsive.

Conclusion
The thoughtful assessment of volume responsiveness in the septic shock patient is an important skill for an emergency physician to possess. “Filling the tank” without concern for the physiologic effects of over-resuscitation may worsen patient outcomes. Cardiac filling pressures (CVP or PAOP) do not perform well at predicting volume responsiveness, and given the multitude of other available options in the emergency physician's armamentarium, alternative methods should be used to predict volume responsiveness.

- - - - -

Questions
1. Assessing fluid responsiveness traditionally involved
     A. Cardiac filling pressures
     B. Central venous pressures
     C. Pulmonary artery occlusion pressures
     D. All of the above

2. The goal of fluid resuscitation in the setting of a critically ill or septic patient is
     A. Increase cardiac demand
     B. Improve cardiac output
     C. Prevent end organ damage
     D. Increase urinary output
     E. B and C

3. The CVP is only slightly better than chance at determining fluid responsiveness.
     A. True
     B. False

4. Ultrasound of the IVC is limited in which type of patient?
     A. Patients with a history of  CHF
     B. Patients with cirrhotic liver disease
     C. Patients with a DVT

5. Passive Leg Raise is useful in situation when
     A. You need to quickly asses volume status
     B. Your patient has cirrhotic liver disease
     C. Your patient has a dysrhythmia
     D. You have not yet gone to the US course in January and do not have a good grasp on cardiac US
     E. All of the above