This talk is going to discuss monitoring of intravascular volume and volume responsiveness. Once again, you can see my credentials and my contact information. And these are my conflicts of interest. I want to stress that any figures that I used here are for illustration only and I'm not advocating for the use of one product or specific class of products at all. The objectives of this talk are to understand how to assess blood volume, mostly how to assess volume responsiveness, and then understand the limitations of both of these assessments. And I want to emphasize that there is an uncomfortable truth about monitoring, and that is that no monitoring device or technique, no matter how accurate or insightful, will improve patient outcome unless it is coupled with treatments that are going to improve outcome. And the big question is whether monitoring something that we think is physiologically important yields clinical benefit to patients. And in a subsequent talk, I'm going to discuss whether these techniques do that. Why do we monitor? Right now in our best model, the main purpose for human endemic monitoring in the ICU is to treat or prevent shock. And in our current understanding, we can think of shock as a condition that occurs when tissues don't receive enough blood flow or oxygen delivery, leading to increased deficit of energy substrates and decreased energy expenditures by cells, causing altered cellular function and then poor organ function. It's a pretty good model, but it appears to have a lot of gaps in it. We sum this up in a relatively recent review paper. You can see the reference there in the lower right, where the ultimate goal of fluid and hemodynamic monitoring is to support normal cellular metabolic function. And achieving normal cellular metabolic function requires maintenance or restoration of effective coordinated function of the macrocirculation and the microcirculation, as well as intact cellular metabolism. Typically speaking, what we do most of the time is targeted at macrocirculatory variables or surrogates of cellular metabolism. And the therapeutic rationale of fluid administration is to optimize macrocirculatory function in order to improve or optimize microcirculatory and therefore cellular function. You could really make an argument that much of intensive care medicine, from mechanical ventilation to intravenous fluids to vasoactive drug infusions, are aimed at manipulations of oxygen delivery. There are several components of tissue oxygen delivery that worry us, and we can manipulate them with therapeutic interventions. We can give red blood cell transfusions if we believe that they increase oxygen carrying capacity. We can increase FiO2 or PEEP or try inhaled nitric oxide or recruitment maneuvers to improve oxygen content. Or we can try to improve the cardiac output. Understanding circulating blood volume is sometimes a pertinent clinical question. Clinical assessment of total circulating blood volume is possible. It's complicated, though. It requires some kind of labeling, either with a dye or radioisotopes. There is a commercially available product that uses radiolabeled albumin, and I've used this product in some research. It can be performed a few or several times a day. The results, however, are not available immediately, and I think we can safely say that this is not really what we think of when we think of monitoring. It's not typically useful for rapid assessments and for understanding patient responses to interventions. And so we're not going to really talk about monitoring of blood volume further. Because the goal of shock resuscitation is to restore blood flow and tissue perfusion, we typically monitor cardiac output and blood pressure. By doing this, we assess the ability of the cardiovascular system to deliver oxygen to tissues and measure the effectiveness of the interventions that we use to try to improve oxygen delivery or tissue perfusion. It's important to keep in mind, however, the limitations of these parameters. Measurements like stroke volume, filling pressures, vascular resistance are so-called macrocirculatory parameters. That is, they help us understand what's going on inside the heart and the large conducting blood vessels. These measurements are important and maybe even necessary, but they don't give us the whole picture. A major limitation in our knowledge is the difficulty or impossibility of measuring tissue perfusion, tissue oxygenation, or tissue oxygen use. So if we define shock as the failure to deliver or utilize adequate amounts of oxygen, we can define it as the oxygen demand, VO2, being greater than oxygen delivery, DO2. And when we think about DO2, we think about cardiac output, Q, and oxygen-carrying content of the arterial blood, CaO2, which is, as you can see, a combination of hemoglobin concentration, oxygen saturation, and a very small contribution of oxygen dissolved in the plasma. We know that low blood pressure is synonymous with shock, but we also know that it doesn't diagnose shock particularly well. Nevertheless, thinking about blood pressure is useful for thinking about shock. It's useful to think about shock in four components. When we think about the blood pressure, we think that the blood pressure is composed of cardiac output and systemic vascular resistance. So shock can result from vasodilation, such as occurs with sepsis, anaphylaxis, or adrenal failure, or it can occur when the cardiac output is compromised, which can happen if the pump is broken, the pump is blocked, or the pump is empty. When the pump is empty, the answer is reasonably straightforward, volume resuscitation. In the other three scenarios, pump block, pump broken, or vasodilation, we have to treat the underlying cause. We can give thrombolytics or do pericardiocentesis. We can open up the coronary artery that's blocked, take the patient to coronary artery bypass surgery, put in a ventricular assist device, or perform a transplant. We can give antibiotics or surgically relieve the site of infection. We can give epinephrine or steroids. Often while we're waiting for these definitive therapies to take place, we give intravenous fluids, pressors, inotropes, put in balloon pumps, or use devices such as the Impella. IV fluids are often the first choice. They're readily available. They're inexpensive. Widely, we perceive them to be benign. Many of the questions that surround their use are which fluid, will it help, and when to stop. That is, how much fluid should we give before it's not safe anymore? In fact, the use of IV fluids is more or less codified in the surviving sepsis guidelines, which tell us to give 30 mLs per kg to patients with sepsis and evidence of poor perfusion. Those guidelines have been in place now for about 20 years. The use of IV fluid in clinical medicine is now almost 200 years old when documentation of use of intravenous fluid to treat severe hypovolemia related to cholera. In the article describing this use, the object of giving IV fluids was to place the patient in nearly his ordinary state as to the quantity of blood circulating in the vessels. Today, we translate that question at the bedside. Where is my patient in relation to the ordinary state? That is, is my patient wet or dry? I would argue that this shorthand is probably not the best way we could phrase this important clinical question, since we can't measure the circulating blood volume easily at the bedside, and also volume status is a relative rather than absolute characteristic. The clinical question about fluids is, rather, whether they will help address the cardiac output. Keeping in mind that shock is a condition in which oxygen and substrate delivery is insufficient to meet tissue demands, we should keep in mind that IV fluids act on the cardiac output, part of the oxygen delivery equation. When we give IV fluids to increase blood pressure, or to increase urine output, or to lower the lactate, what we're really saying is that we're trying to increase the blood pressure or renal perfusion or tissue perfusion by increasing cardiac output. So instead of asking one question, is my patient wet or dry, we're really asking several questions. If I give IV fluid, does the cardiac output increase by increased stroke volume? And if I give fluids to increase the cardiac output, does that increase oxygen delivery? Does oxygen delivery lead to improved organ function? Unfortunately, we don't have really good ways of understanding the final two questions in this series of questions, so we kind of focus on the first three questions. Other questions we ought to be asking at the same time is, giving fluid harmful? What kind of fluid should I give? Isotonic crystalloid, colloid, blood components, and when do we stop? We have to keep in mind that fluids have possible adverse effects, including pulmonary edema and inflammation, interstitial kidney and intestinal edema, elevated abdominal pressure, and disruption of the endothelial glycocalyx. Traditional hemodynamic monitoring gives us information about filling pressures, such as CVP or pulmonary artery occlusion pressures. We can also give us information about oxyhemoglobin saturation in the mixed venous blood. This allows us to assess oxygen extraction and calculate the cardiac output using the Fick equation. However, these traditional static hemodynamic measurements, such as CVP and wedge pressure, are unreliable ways to answer some practical bedside questions. It's important to keep in mind other limitations. The coefficient of variance of thermodilution cardiac output measurements is high, although it narrows when you take respiratory variation into account. And the modified Fick method relies on various assumptions, some of which are unreliable in critically ill patients. The pulmonary artery occlusion pressure, also known as the wedge pressure, and the central venous pressure unfortunately do not help us identify patients who are likely to be volume responsive easily. In my next talk, I'm going to discuss volume responsiveness more in depth. There are measurements that help obtain data relatively easily at the bedside, and they appear to have reasonable ability to predict whether an individual's patient's stroke volume, cardiac output, and presumably tissue perfusion will improve when we infuse intravenous fluids. These methods are the stroke volume variation and the pulse pressure variation, both of which have been shown in multiple studies to have good ability to predict volume responsiveness. This is just one example of those kinds of papers. This topic has been extensively reviewed. It's important to keep in mind that an individual patient's heart may be operating on one of many Frank-Starling curves. Parameters such as pulse pressure variation or stroke volume variation help us understand which part of a Frank-Starling curve best describes a given patient's heart at a given moment when we're taking the measurements. So the core question involved in functional hemodynamic monitoring is, at this moment, is the patient on the steep part of the curve or the flat part? No static test, CVP, wedge pressure can tell you which curve represents a patient's cardiac performance at a specific time or which part of that curve describes the patient's heart at that specific time. Let's remember that the indication for giving IV fluids for shock is to increase left ventricular stroke volume. IV fluid loading will only increase stroke volume if ventricular function is on the steep part of the Frank-Starling curve. If a patient's heart is operating on the vertical part of the curve, fluid infusion is likely to increase stroke volume. If the patient's heart is on the more horizontal or flat part of the curve, fluid is very unlikely to increase stroke volume significantly. Please keep in mind that to assess volume responsiveness, you really need to measure a stroke volume and cardiac output directly. Using measurements such as blood pressure, changes in pulse pressure variation, and so forth are not good ways of understanding fluid responsiveness. There are a lot of products available on the market today that will help understand volume responsiveness. A standard A-line may be able to do it if you can accurately identify inspiratory and expiratory phases of respiration. And there are a variety of other minimally invasive or even non-invasive ways to measure a cardiac output, as you can see. The most accurate methods of measuring cardiac output use calibration. Essentially, you have to calibrate one of these catheters by putting in a special arterial catheter that has an indicator built into it, and that indicator might be cold fluid or lithium. That indicator is injected into the central vein and then detected by the arterial catheter. The stroke volume and cardiac output measured by the indicator dilution technique is then compared to the pressure waveform, and changes in the pressure waveform over time are interpreted as changes in stroke volume and cardiac output. So you can see how injection of an indicator, whether it be dye or cold saline, gives information based on stroke volume, or the slope of the decay curve can give information about how much thermal loss goes into the lungs, which is an estimate of extravascular fluid leakage. There are many forms of uncalibrated monitors available too, FlowTrack, ProAct, Lidco Rapid. I put an asterisk on ProAct because it's not currently available in the United States. All of these devices analyze pressure waveforms and calculate the stroke volume and cardiac output based on algorithms that incorporate a bunch of assumptions based on patient age, sex, previous medical history, and so forth. All of these devices have some inaccuracy in their readings of stroke volume and cardiac output, but their linearity is very good. That is, over time they will generate a cardiac output or stroke volume curve that is parallel to but deviates from the true line that describes cardiac output over time. As an illustration, the red line might describe the true cardiac output after an epinephrine infusion is given. The blue lines describe cardiac output as it is read by various uncalibrated devices. Now this is useful for tracking response to treatment, but it might be less useful when making a decision to start treatment or stop treatment. For example, if your shock patient has a cardiac output of 3, but the monitor says 1 or 6, you might have very different ideas of what the correct intervention would be. Uncalibrated waveform devices, such as PROACT or FlowTrack, use interpretation of the waveform and then proprietary algorithms to estimate the cardiac output, as I discussed previously. This is just an illustration of which part of the pressure-time curve they use to make those estimates of stroke volume and cardiac output. There are several non-invasive cardiac output monitors, all of which are uncalibrated. The Starling SV device, which is shown on the left, uses bioimpedance bioreactants. Finger cuff blood pressure devices, such as ClearSight or CNAP, use changes in arteriolar blood flow and arterial blood flow in the digits to make estimates of stroke volume and cardiac output. There are some Doppler-based cardiac output monitors that are also uncalibrated. The USCOM device, the Deltex device, and a newer device called Flosonics Flowpatch all look at Doppler flow in the central arteries to make estimates of the stroke volume. As we discussed, in the last 20 years, the study of functional hemodynamic monitoring has emerged. This kind of monitoring takes advantage of dynamic measurements of stroke volume in response to changing preload conditions to help us understand which part of the Frank-Starling curve describes an individual's heart, and thus whether that individual will show a higher stroke volume with IV fluid loading. Many of the underlying principles are difficult for some people to understand, but they're based on the same principles as the Pulsus Paradoxus that we learned about in medical school, just as we discussed in the previous talk on heart-lung interactions. The classic fundamental hemodynamic monitoring maneuver is to examine variations in arterial pressure waveforms in response to changes in intrathoracic pressure and lung volume that result from mechanical ventilation. Again, this is related to Pulsus Paradoxus. When a positive pressure breath is administered, intrathoracic pressures and lung volumes rise, leading to a decrease in venous return and a decreased stroke volume a few heartbeats later. Importantly, note how these arterial pressure devices estimate the stroke volume. They look at the area under the curve of the arterial pressure time waveform. In a calibrated device, the area is compared to the thermodilution technique or the indicator dilution technique to establish a value for the stroke volume. In uncalibrated devices, the area under the curve is compared with an algorithm to give the value for stroke volume. This graph is from Shel Magner's 2004 review article in the American Journal of Respiratory and Critical Care Medicine. There were no spontaneous breathing efforts, and the patient is receiving volume control ventilation with a tidal volume of about 700 milliliters and a rate of 12 per minute. The delta up and delta down components of spontaneous variations are shown. The bars at the bottom mark inspiration. The arrow at the point mark apnea represents an expiration value for determining delta up and delta down. These findings also hold somewhat less accurately with lower tidal volume ventilation. Using the pulse pressure variation technique, the maximal difference in arterial pulse pressure between end inspiration and end expiration is measured. The principle here is that pulse pressure variation is greater on the lower, steeper part of the Frank-Starling curve. A pulse pressure variation of greater than 13% has a very accurate positive predictive value for identifying volume responsiveness in a broad variety of ICU patients. In stroke volume variation, you need to use one of the devices that actually measures stroke volume based on analysis of the blood pressure time curve. The principle here is that stroke volume variation is greater on the lower, steeper part of the Frank-Starling curve. A stroke volume variation of greater than 10% accurately predicts volume responsiveness in a broad variety of ICU patients. Similarly, we can use the passive leg raise maneuver. In this maneuver, we use the body's own reservoir of about 500 milliliters of blood that rests in the leg veins to test whether cardiac output will likely improve with a volume infusion. We rapidly change the patient from a semi-recumbent position to a flat with legs raised 45 degrees position. Changes in stroke volume when the leg raises perform identifies patients who will respond to fluid. It's important to recognize that there are limitations on when pulse pressure variation and stroke volume variation are accurate. Patients must be receiving positive pressure ventilation without spontaneously triggered breaths. They cannot have atrial arrhythmias like atrial flutter, atrial fibrillation, and so forth. It can be difficult to do at the bedside without printing out tracings or doing intervals, which we don't really do at the bedside. But it will work when you have specialized equipment such as some of the devices that we discussed previously. It's also important to keep in mind that a passive leg raise will work in these situations. So the patient can have atrial arrhythmias, can be triggering breaths, does not need to be on positive pressure ventilation for a passive leg raise to work. But you still need to measure stroke volume and cardiac output. Passive leg raising will not work in patients who have a high intra-abdominal pressure because the high intra-abdominal pressure tends to dampen the effect of blood infusion from the legs into the central veins. Other maneuvers will work in these conditions such as the end occlusion maneuvers or responses between drastic changes in PEEP. Again, you need to measure stroke volume and cardiac output to read out these principles. Bedside ultrasound is believed by many to show some promise as a test that helps us answer these fluid questions, but it has many, many, many important limitations. First, it's not a continuous monitor. Second, there is a great deal of inter-observer variability in reading the images. Third, it requires rather precise measurements of the IVC diameter, which will depend on the spot at which the measurement is taken, that is, where is the probe placed. If all of the images are probe placed, if all of these elements are not consistent, then the measurements will have poor characteristics as a diagnostic test. And the limits of ultrasound are discussed in this paper, which is fun and highly worth reading. In the subsequent talk, we're going to discuss how to incorporate these monitoring techniques into making decisions about giving intravenous fluids. Thanks for your attention.

intravascular volume

volume responsiveness

shock treatment

monitoring devices

stroke volume variation

fluid administration