Fluid Responsiveness, “Answers”

1. How do you assess fluid responsiveness in the ED? Do you use IVC collapsibility in spontaneously breathing patients?

Although fluid resuscitation is paramount in the treatment of sepsis, volume overloading critically ill patients has been shown to worsen outcomes including length of intensive care unit (ICU) stay, days on a ventilator, and mortality (Rosenberg, 2009). Methods of assessing volume status (preload) and hemodynamic response to fluid challenges (volume responsiveness) are thus very important when managing these patients. Until recently, central venous pressure (CVP) monitoring dominated clinician guidance of fluid management and was used regularly by over 90% of intensivists (McIntyre, 2007). CVP represents the right atrial pressure and has erroneously been extrapolated to estimate left ventricular preload and thus fluid responsiveness. A recent meta-analysis found no relationship between CVP and circulating blood volume, left or right ventricular preload, or fluid responsiveness (also known as the “seven mares” article, Marik, 2008). Alternative methods of determining volume status and fluid responsiveness have subsequently been sought with greater fervor.

EML fluid responsiveness answers

Pulmonary artery occlusion pressure (PAOP) measured via a pulmonary artery catheter, like CVP, fails to reflect preload or volume responsiveness (Marik, 2010). Other static indices including left ventricular end-diastolic area (LVEDA) measured by transesophageal echocardiography and global end-diastolic volume measured through a cardiac output monitor (PiCCO), although predictive of preload, also fail to accurately predict fluid responsiveness (Marik, 2010).

Dynamic measurements perform better in predicting fluid responsiveness but generally require mechanical ventilation to control for substantial variation in respiratory cycle volumes and intrathoracic pressures characteristic of spontaneous breathing patterns. Pulse pressure variation (PPV) measured by arterial waveform and stroke volume variation (SVV) measured by arterial or pulse oximeter plethysmographic waveform, have been show to correlate very well with volume responsiveness. The sensitivity and specificity of PPV has been documented at 89% and 88% respectively, and that of SVV has been documented at 82% and 86% (Marik, 2010). Accurate measurements do require tidal volumes of 8-10L/kg and specialized analysis devices. Inferior vena cava diameter distensibility (dIVC) with respiration, although criticized by some as having limitations similar to CVP (Marik, 2010), has been studied repeatedly in mechanically ventilated patients and appears to be a valid option for predicting volume responsiveness. Barbier showed that dIVC >18% predicts volume responsiveness with a sensitivity and specificity of 90% (Barbier, 2004). Other studies, though small and observational, show a similar correlation (Machare-Delgado, 2011; Moretti, 2010).

Unfortunately, all the aforementioned techniques possess limitations that will often preclude their application in the Emergency Department. In the ED, the ideal method for measuring fluid responsiveness must be technically easy, fast, non-invasive and, importantly, reliable in spontaneously breathing patients. SVV measured by arterial waveform has been shown to be predictive of volume responsiveness in spontaneously breathing patients at a threshold of 17% (PPV 100%, NPV 82%, p=0.03) in at least one study (Lanspa, 2013). This technology, however, requires placement of an arterial line and specialized equipment not available in most EDs. Similarly, straight leg raise predicts fluid responsiveness reliably but requires invasive monitoring like an a-line or specialized equipment (Benomar, 2010). IVC collapsibility on the other hand is technically easy, non-invasive and recent studies suggest it may have a role in spontaneously breathing patients. In spontaneously breathing patients, the IVC collapses on inspiration and distends on expiration.  Upon intubation, the patient’s physiology reverses from negative pressure to positive pressure.  As a result, the IVC distends on inspiration and collapses on expiration. The best available IVC data consists of two observational studies, which ultimately offer cautious support for use of IVC collapsibility in breathing patients. The first found IVC inspiratory variation greater than 40% to predict fluid responders with a sensitivity of 70% and specificity of 80%. Values of below 40%, however, could not be used to exclude fluid responders (Muller, 2012). The second study found variations in inferior vena cava index less than 15% to have 100% negative predictive value (p=0.03) for fluid responsiveness whereas over 50% variation had a positive predictive value of 75% (p=0.09) (Lanspa, 2013). Both studies used subcostal windows to assess inferior vena cava diameter variation as it entered the right atrium. Though promising, this data should be interpreted carefully given the small size of the studies, the lack of statistical significance for some values, and the wide range of clinically indeterminate values of IVC collapsibility.

It is important to remember that all of the cited studies apply to initial resuscitation in the ICU, often after aggressive fluid resuscitation in the ED. The need for a more cautious approach to fluid resuscitation during the initial management of critically ill, particularly septic, patients in the ED is less established.

2. Which crystalloid fluid do you use to resuscitate critically ill patients?

Normal saline (NS) is traditionally the first-line fluid for resuscitation of critically ill patients in the ED. NS first came into widespread use in the 1830’s during the European cholera epidemic, saving countless lives. The actual electrolyte content of NS during its early days was likely more “normal” than it is today, with estimated levels of sodium and chloride at 134 and 118 mmol/L respectively (Yunos, 2010). Today, NS is neither normal nor physiologic, containing 154 mmol/L of both sodium and chloride. Every liter of NS administered thus delivers supra-physiologic levels of these electrolytes, which play key roles in the acid-base physiology. Alternate crystalloid solutions including Hartmann’s Lactated Ringers (LR) and balanced electrolyte solutions (BES) such as Plasma-Lyte offer more physiologic concentrations of electrolytes and may have unique advantages for resuscitation in critical care. Small variation in electrolyte content can make clinically important differences when resuscitating with large volumes or when caring for patients over extended periods in the ICU. In the current era of hospital overcrowding and extended ED stays, this concern becomes particularly relevant to all ED physicians.  See table below for details of electrolyte content of commonly used fluids (Table 1).

Table 1: Electrolyte Content of Common Crystalloid Solutions (mmol/L)

Plasma Normal Saline Hartmann’s LR Plasma-Lyte
Sodium 140 154 131 140
Potassium 5 0 5 5
Chloride 100 154 111 98
Bicarbonate 24 0 0 0
Calcium 2.2 0 2 0
Magnesium 1 0 1 1.5
Lactate 1 0 29 0
Acetate 0 0 0 27
Gluconate 0 0 0 23

Specifically, high chloride content has been targeted as a potential source of harm in large volume crystalloid resuscitation. New understanding of complex acid-base physiology, namely the Stewart physiocochemical approach, is the driving force behind recent attention given to chloride. Briefly, under this approach, chloride is the predominant negative strong ion in plasma and a key component of the strong ion difference (SID), which directly influences hydrogen ion concentration and thus acid base status (Yunos 2010). NS resuscitation has been clearly linked to hyperchloremic metabolic acidosis (HMA), but debate exists regarding its clinical significance (Yunos, 2010; Heijden, 2012). Preclinical and healthy human volunteer data provide increasing evidence for chloride-associated hypotension, reductions in renal cortical perfusion, decreased glomerular filtration rate (GRF) and pro-inflammatory states (Chowdhury, 2012; Yunos, 2010; Wilcox, 1983; Kellum, 2004; Kellum, 2006). Recently, a prospective, open-label study looked at chloride liberal vs. chloride restrictive fluid resuscitation of critically ill patients and its effect on acute kidney injury (AKI). Importantly, in this study of over 1500 patients, resuscitation with chloride restrictive fluids was associated with statistically significant lower rise in serum creatinine levels and less incidence of AKI. Though a secondary outcome, patients receiving chloride restrictive fluids also received less renal replacement therapy (Yunos, 2012). The combined existing evidence, now bolstered by a well-designed clinical trial, calls into question the routine use of potentially harmful chloride-rich fluids when alternative, equally effective options are available.

Choice of crystalloid fluid may be particularly important in conditions with disarray of electrolytes and acid-base status such as diabetic ketoacidosis (DKA). Patients in DKA are profoundly volume depleted and require large volumes of NS for resuscitation. As a result, HMA commonly occurs during treatment and complicates the management of DKA (Morgan, 2002). A blinded, randomized controlled trial compared a balanced electrolyte solution (BES), Plasma-Lyte, to NS for prevention of HMA during resuscitation of patients with DKA. Patients receiving BES were found to have significantly lower levels of chloride and higher levels of bicarbonate, consistent with prevention of HMA (Mahler, 2010). A smaller, non-randomized study found similar results (Chua, 2012). Less evidence is available for LR and DKA. A randomized controlled trial compared NS to LR for resolution of acidosis. This study was small and terminated early due to poor enrollment; there was a non-significant decrease in time to resolution of acidosis in the group receiving LR (Van Zyl, 2011) As mentioned previously, the clinical significance of HMA is still debated, but mounting evidence suggests avoidance of HMA may be beneficial to the patient.

3. Do you ever use hypertonic saline in patients with septic shock?

Through multiple inflammatory mechanisms, sepsis creates a pathophysiologic state of vasodilation and increased endothelial permeability with resultant maldistribution of blood flow. Rapid and high-volume fluid resuscitation is a key element to counter this effect and to adequately deliver oxygen to tissues in patients with septic shock. Hypertonic fluids may offer unique benefits over other crystalloids. Hypertonic saline osmotically pulls fluid from intracellular spaces into the vasculature, resulting in rapid plasma expansion that supersedes the actual volume infused. This effect permits use of smaller fluid volumes, decreasing risk of edema, further improving oxygenation of tissues. Preclinical data supports the use of hypertonics in sepsis, with cardiovascular benefits ranging from improved volume expansion to increased cardiac contractility and better splanchnic perfusion (Garrido, 2006; Ing, 1994; Oi, 2000). Additionally, enhanced immunomodulatory effects including reduced bacterial colony counts and enhanced bacterial killing have been demonstrated with hypertonics (Shields, 2003).

Good clinical data on hypertonics and sepsis, however, is limited and further studies are needed. Two small, randomized controlled trials evaluated an initial bolus of hypertonic saline with colloid compared to colloid or NS alone and found improved cardiac function with hypertonics (Oliveira, 2002; van Haren, 2012). In Oliveira’s study, the group receiving 7.5% saline/dextran was found to have significant increases in cardiac index, pulmonary artery occlusion pressure and stroke volume index without significant side effects (Oliveira, 2002). Van Haren found the 7.2% hypertonic/hydroxyethyl starch (HES) group to have increased cardiac contractility and a decreased need for further fluid resuscitation in the following 24 hours. Although these studies were randomized, both were extremely small thus preventing the evaluation of clinically important measures including mortality and potential risks including hypernatremia and acid-base effects.


4. What is your threshold for giving blood transfusions? Does this change in patients with cardiac disease or GI bleeds?

In states of high metabolic demand accompanying critical illness, oxygen requirements can outpace supply, creating an oxygen debt at the tissue level. Allogeneic red blood cell (RBC) transfusions have long been a cornerstone in critical care management to counter this imbalance and augment delivery of oxygen to tissues. Prior to the TRICC trial in 1999, a hemoglobin (Hgb) transfusion threshold of 10 g/dL was standard practice. Growing concern over the complications of RBC transfusions, including immunosuppression, inflammation, infection and transfusion reactions, particularly in the critically ill, prompted the landmark TRICC trial. This was a randomized controlled trial comparing a restrictive versus liberal (7.0 g/dL vs.10.0 g/dL) Hgb transfusion threshold. Actively bleeding patients and those with acute coronary syndrome (ACS) were excluded; patients with cardiac disease were included. The TRICC trial showed no difference in 30-day mortality for a restrictive compared to liberal transfusion threshold (18.7% vs. 23.3%, p=0.11). Additionally, fewer cardiac adverse events and smaller changes in multi-organ system dysfunction scores were seen in the restrictive group (Hebert, 1999). This trial firmly established a threshold of 7.0 g/dL as an acceptable Hgb transfusion strategy in the critically ill. Supporting this conclusion, a 2012 Cochrane review found restrictive strategies to result in a 39% reduction in blood transfused, an overall reduction in in-hospital mortality, and no difference in mortality at 30 days (Carson, 2012).

In a subgroup analysis of the TRICC trial, the restrictive arm showed no difference in 30 and 60-day mortality for patients with cardiovascular disease 20.5% vs. 22.9% (p=0.69). This finding differed significantly from preexisting observational data, which showed increased mortality with a restrictive strategy (Carson, 1996). Complicating the picture, when confirmed ischemic heart disease, severe peripheral vascular disease, and severe comorbid cardiac disease were isolated from all cardiac disease (i.e. group of most clinically relevant cardiac disease), a non-significant trend towards increased mortality was seen in the restrictive group (p=0.3) (Hebert, 2001).

To address this discrepancy, the FOCUS trial compared a liberal (Hgb <10 g/dL) vs. restrictive (Hgb <8 g/dL or symptomatic) transfusion threshold in patients with CAD or CAD risk factors undergoing hip surgery. Using a composite endpoint of death and inability to walk independently, the restrictive strategy was found to be no different. No difference was found in secondary outcomes of adverse cardiovascular events (Carson, 2011). This study was billed to be the definitive trial for restrictive transfusion thresholds in patients with CAD but it has received significant criticism. Utilization of a composite end point with components differing greatly on clinical significance (walking independently and death) can cloud results and lead to misleading interpretations. Although mortality was reduced in the restrictive strategy (6.6% vs. 7.6%), a much larger sample size would be required to draw significant conclusions (Meybohm, 2012). The American Association of Blood Banks’ (AABB) clinical practice guideline offers a weak recommendation for transfusion of hemodynamically stable patients with cardiovascular disease at Hgb concentrations of 8 g/dL or for symptoms (Carson, Grossman, 2012).

To date, no randomized controlled trial of transfusion strategies in patients with active ACS has been undertaken. A review of existing studies consisting primarily of observational data concluded that in patients admitted for ACS, transfusions at Hgb >11 g/dL increased mortality but at Hgb <8 g/dL, transfusions decreased mortality or did no harm. Given the observational nature of the studies, however, conclusions cannot be drawn (Garfinkle, 2013). The AABB does not make a recommendation for transfusion thresholds in patients with ACS, citing absence of quality data (Carson, Grossman, 2012).

A restrictive transfusion strategy appears to be safe in patients with CAD, but importantly, none of the above trials included actively bleeding patients. In 2013, Villanueva published a landmark paper in the New England Journal addressing transfusion thresholds in patients with acute upper GI bleeds. In this trial patients were randomized to transfusion Hgb thresholds of 7 g/dL vs. 9 g/dL. Patients in the restrictive group had significantly decreased bleeding, fewer adverse events and increased survival at 6 weeks. With this evidence, patients with active upper GI bleeds can now be considered prime candidates for restrictive transfusion thresholds, which may not only be safe, but beneficial (Villanueva, 2013).

This entry was posted in Uncategorized and tagged , , , , . Bookmark the permalink.

5 Responses to Fluid Responsiveness, “Answers”

  1. Pingback: electrolyte content of crystalloids (quick reference) | DAILYEM

  2. Great, great, great, great….GREAT post! Excellent pooled sources and info! Keep ’em coming!

    I was disappointed, however, to see that you did not mention PLR plus ETCO2. The study below (although small) suggested that a PLR-induced increase in ETCO2 ≥ 5 % predicted a fluid-induced increase in cardiac index ≥ 15 % with sensitivity of 71 % (95 % confidence interval: 48-89 %) and specificity of 100 (82-100) % — In my opinion, this is one of the more easily performed procedures & can be completed quite easily in the ED if the patient is on a ventilator.

    What are your thoughts on this?

    Monnet et. al (2013). “End-tidal carbon dioxide is better than arterial pressure for predicting volume responsiveness by the passive leg raising test.” Intensive care medicine, 39(1), 93–100. doi:10.1007/s00134-012-2693-y

    • Anand Swaminathan says:

      Derek, sorry for the delayed response. We wrote these “answers” before this article came out (we have a delay to posting to the website). I just pulled the article you referred to and will review it and get back to you. Thanks!

    • Anand Swaminathan says:

      Derek, thanks again for the article. Very interesting and would be great if it works as it’ll be another arrow in the non-invasive monitoring quiver. I would like it more if the study was done in the ED instead of in the ICU. It clearly also needs to be replicated with a larger n. The other thing was that they found that a 5% difference in EtCO2 was present but the range was 0 – 13% which suggests to me that many patients had no change. A 5% change is really modest and may be difficult to appreciate. The theory is interesting but I think I would need more for it to be ready for prime time.
      Thanks again for sending our way and please let us know if you are doing this.

  3. Pingback: Fluid Responsiveness: EM Lyceum | The KP EM Residency

Comments are closed.