Femoral-Based Central Venous Oxygen Saturation Is Not a Reliable Substitute for Subclavian/Internal Jugular-Based Central Venous Oxygen Saturation in Patients Who Are Critically Ill FREE TO VIEW

In the setting of shock, an imbalance between the metabolic demands of the body and adequate delivery of oxygen to its tissues often leads to a cascade of events, including anaerobic metabolism and mitochondrial dysfunction, culminating in multiorgan system failure and death.^1-3 Rapid identification and intervention based on markers of tissue hypoperfusion and dysoxia are, therefore, essential to the treatment of shock and may improve outcomes.^1,4,5

One indicator that can reflect the relationship between global oxygen delivery and demand is the mixed venous oxygen saturation (Svo₂), obtained from the distal port of a pulmonary artery catheter (PAC). When oxygen delivery has been compromised or oxygen consumption has exceeded its supply, subsequent oxygen venous return to the right side of the heart is diminished. An Svo₂ < 65% (normal Svo₂: 65%-75%) reflects this imbalance and has been shown to occur in the setting of normal vital signs and adequate urine output.^1,5-9 Although Svo₂ detects real-time imbalances between systemic oxygen delivery and demand, recent studies suggest that when obtained early in the disease process, timely recognition and intervention lead to improved outcomes.^5,10 Given the time-sensitive nature of obtaining this value and the logistic challenges of inserting a PAC in every patient, the central venous oxygen saturation (Scvo₂) (superior vena cava) has largely replaced the PAC-derived Svo₂.^1,4,11 This strategy of immediate and appropriate resuscitation using hemodynamic variables and Scvo₂ was demonstrated in the trial of early goal-directed therapy by Rivers et al.⁵ Although Svo₂ and Scvo₂ are not equivalent values, they have been shown to trend throughout various physiologic states.^12-20 A low Scvo₂ is clinically relevant. It can be used as a surrogate for Svo₂ and as an end point in resuscitation to improve survival.^{5,7,11-13,19,21-27}

Although multiple studies suggest that early, goal-directed care improves outcomes, widespread implementation of this strategy has not yet been achieved.^28-33 In order to monitor the Scvo₂ during resuscitation, an internal jugular or subclavian line must be inserted immediately upon patient presentation. However, it has been our experience that the femoral vein is frequently the preferred site for initial access when caring for patients who are critically ill, particularly in the ED. We have also observed that measurement of femoral venous oxygen saturation (Sfvo₂) is sometimes used as a surrogate for Scvo₂ to help guide resuscitation_. However, it is unknown if Sfvo₂ is equivalent to, or even consistently trends with Scvo₂. The purpose of our study is to determine the frequency of femoral central line use in the resuscitation of patients who are critically ill and to determine if samples of Scvo₂ and Sfvo₂ can be used interchangeably in a population of patients who are critically ill.

We conducted a survey (Fig 1) of attending physicians who care for patients who are critically ill at medical centers throughout the United States. We used an Internet-based survey tool known as SurveyMonkey to solicit responses (http://www.surveymonkey.com) (SurveyMonkey; Portland, OR).³⁴ The SurveyMonkey questionnaire was administered via e-mail to physician members of professional societies, including the Society of Critical Care Medicine, the American College of Emergency Physicians, and the American Society of Anesthesiologists. Responses were compiled and analyzed by the SurveyMonkey program. No identifiable material was collected, and the respondents’ anonymity was maintained.

A single-center, prospective study was conducted at The George Washington University Hospital’s critical care unit. This ICU is a closed, level I, 48-bed, combined medical-surgical unit that admits all adults who are critically ill, except those with major thermal injuries and solid organ transplants. The George Washington University Medical Center Review Board approved the study. Informed consent to participate in the study was obtained from patients or their next of kin.

Patients were screened for inclusion and exclusion criteria between June 2007 and May 2008. Inclusion criteria included nonpregnant adults (≥ 18 years old) who had both a femoral venous catheter and a nonfemoral central venous catheter inserted at any point in their ICU care. Femoral catheters were all 20 cm in length (Edwards Lifesciences Corporation; Irvine, CA, and AngioDynamics, Inc; Queensbury, NY). The nonfemoral venous catheters included subclavian, internal jugular, or peripherally inserted central catheters (Edwards Lifesciences Corporation and C. R. Bard, Inc; Murray Hill, NJ). For example, a femoral catheter was placed in a patient who needed urgent medical therapy in the ED and could not wait for radiographic confirmation. After stabilization in the ICU, a new, sterile subclavian line was inserted, thus enabling temporary access to both femoral and nonfemoral venous blood. All central venous catheters were confirmed to be in the distal portion of the superior vena cava by chest radiograph. Physicians not involved in the study determined the need for and placement of the catheters as part of standard ICU care.

Blood samples were drawn simultaneously from the distal ports of the femoral and nonfemoral central venous catheters at times zero and 30 min. The first 5 mL of blood drawn from each sample was discarded to prevent dilution. Blood was immediately placed on ice and evaluated for oxygen saturation and lactate levels using standard blood gas analysis (ABL 700; Radiometer America Inc; Westlake, OH). All venous oxygen saturations were measured by direct cooximetry. Demographic, clinical, and severity-of-illness data were collected and recorded at the time blood samples were drawn. Severity of illness was assessed by use of Acute Physiology and Chronic Health Evaluation (APACHE) II scores and Sequential Organ Failure Assessment scores.^35,36

Scvo₂ and Sfvo₂ values and lactate levels at times zero (T0) and 30 min (T30) were averaged and analyzed individually and in combination. The values were tested for skewedness and kurtosis to assess if the data were parametric or nonparametric. Data were compared by t testing and Wilcoxon analysis as indicated. The Scvo₂ and Sfvo₂ were also assessed by Bland and Altman analyses.³⁷ Scvo₂ and Sfvo₂ were compared with Pearson correlations. The systematic error (bias) and the 95% limits of agreement (mean bias ± 2 × SD) for Scvo₂ and Sfvo₂ were calculated. Bias was expressed as the mean difference of the individual values. This study defined the limits of agreement to be considered clinically acceptable if they were within 5%. Differences > 5%, especially if inconsistent, might prompt inappropriate therapeutic interventions based on a narrow target of resuscitation goals. Statistical analysis was performed by SPSS 11.0 software (SPSS Inc; Chicago IL).

Eight hundred physicians in total were surveyed; 150 (19%) responded. Responses to questions 1, 2, and 3 are provided in Table 1. According to this survey, > 35% of the surveyed physicians insert a femoral line at least 10% of the time during the initial treatment of patients who were critically ill (Fig 2). Various reasons were given for placement of a femoral line. “No other access” was the most frequent response given (Fig 3).

The demographics and clinical characteristics of the 39 patients who were critically ill and enrolled in this study are shown in Table 2. Vasoactive infusions and ventilatory settings were not adjusted between T0 and T30. Similarly, there were no additional fluid boluses given during this time period. Thirty-nine pairs of simultaneously drawn Scvo₂ and Sfvo₂ and nonfemoral lactate and femoral lactate values were recorded at T0 and T30. All mean values were averaged together at T0 and T30. Results can be found in Table 3. Values at T0 and T30 were also analyzed separately. Once again, the difference between Scvo₂ and Sfvo₂ values was statistically significant (data not shown). According to the Bland and Altman analyses (Fig 4), the mean bias between Scvo₂ and Sfvo₂ was 4.0% ± 11.2% and the 95% limits of agreement were large (−18.4% to 26.4%). Therefore > 50% of Scvo₂ and Sfvo₂ diverged by > 5%. Figure 5 shows a correlation between Scvo₂ and Sfvo₂ (r² = 0.35, P = .01).

The aim of the present study was to test the hypothesis that Scvo₂ and Sfvo₂ are tightly correlated and therefore can be used interchangeably. As expected, the Scvo₂ and Sfvo₂ correlated with one another (r² = 0.35, P = .01). Although this value is statistically significant, it does not mean that the tests are interchangeable. In fact, our results suggest that Sfvo₂ is not always a reliable substitute for Scvo₂. While lactate levels did not significantly differ between central and femoral venous sources (2.84 ± 4.0 vs 2.72 ± 3.2; P = .146), mean Scvo₂ and Sfvo₂ values were significantly different (73.1% ± 11.6% vs 69.1% ± 12.9%; P = .002). Although the mean bias between Scvo₂ and Sfvo₂ was low, the SD and therefore the limits of agreement were substantial (4.0 ± 11.2%; 95% limits of agreement: −18.4% to 26.4%). By way of example, according to our results, a Scvo₂ of 70% corresponds to a Sfvo₂ of 66%. Yet when incorporating the sizeable SD, the Sfvo₂ values range from 58.8% to 81.2%. This has considerable clinical implications because the lower values could lead to the unnecessary initiation of inotropic agents or blood transfusions. In fact, some individual Sfvo₂ values differed by > 15% from corresponding nonfemoral central venous values. This is visually depicted on the Bland and Altman projection. which again demonstrated that > 50% of Scvo₂ and Sfvo₂ diverged by > 5% (Fig 4).

Few trials have previously investigated the utility of Sfvo₂. Emerman et al³⁸ analyzed blood gases from the pulmonary artery, central vein, and femoral veins in dogs that experienced cardiac arrest. Those authors reported no significant differences in the oxygen saturation levels, regardless of the venous source. A few human trials have compared Svo₂ with hepatic venous oxygen saturation. Landow et al³⁹ evaluated markers of splanchnic perfusion in patients who had cardiopulmonary bypasses. Svo₂ and hepatic venous oxygen saturation were noted to be significantly different. Similarly, Ruokonen et al⁴⁰ investigated Svo₂ and hepatic venous oxygen saturations before and after vasoactive treatment. Those authors found that while Svo₂ and hepatic venous oxygen saturation trended in a similar direction in response to changes induced by vasoactive infusions, the absolute values were significantly different. Another study reported a 15% mean difference between Svo₂ and hepatic venous oxygen saturation in patients who were septic, but found equivalence in postoperative patients.⁴¹ Our study uniquely compares the relationship between oxygen saturation obtained from the superior vena cava and the femoral vein in a population of patients who were critically ill.

Although little is known about the role for Sfvo₂ in resuscitation protocols, there still exists a need to identify all patients with tissue dysoxia and shock.⁴² The literature to date has largely focused on lactate, Svo₂, and Scvo₂ as the key clinical markers for guiding resuscitation efforts. We have previously shown that Scvo₂ and Svo₂ are not identical, but that these values are well correlated.⁴³ This correlation has been confirmed in multiple studies.^{1,4,5,11-20,23,25} Based on these data, the Surviving Sepsis Campaign has recommended that the goals of initial resuscitation in sepsis-induced hypoperfusion include an Svo₂ of 65% or an Scvo₂ of 70%.⁴⁴

Despite the evidence supporting the use of Scvo₂ as a goal for resuscitative efforts, widespread adoption of this measure has not occurred.^26,28,29 Internal jugular or subclavian venous catheter placement is not infrequently a barrier to protocol implementation in the ED.^29,45 Our survey showed that a significant number of femoral venous catheters are placed during the initial care of patients who are critically ill. Femoral catheter insertion is a quick, safe, and reliable method of obtaining central venous access, especially for patients who need urgent vasoactive infusions and fluid or blood resuscitation. The femoral catheter does not require radiographic verification of position.⁴⁶ Another reason for preferential insertion of a femoral venous catheter in the acute setting is the lack of risk for pneumothorax. Additionally, as more implantable cardiac defibrillators and tunneled catheters are being placed, subclavian and internal jugular access becomes increasingly limited. These reasons were all cited by physicians in our survey who use the femoral vein as the initial site for catheter access. Given the importance of recognizing early tissue hypoxia and the frequency in which femoral venous access is obtained, our study hoped to demonstrate a clinically useful correlation between venous oxygen saturation obtained from a femoral catheter and that drawn from a catheter in the superior vena cava. Yet based on the above data, we conclude that Sfvo₂ is not a reliable substitute for Scvo₂ in goal-directed resuscitation.

The potential reasons for these findings and the underlying pathophysiologic aspects are numerous. First, at a length of 20 cm in a normal-sized adult, a femoral catheter remains in the iliac vein. In this position, the distal tip does not sample the venous return of any intraabdominal organs. Second, in times of physiologic stress, perfusion to kidneys, muscle, and splanchnic regions of the body may be decreased, while flow to the myocardium and brain is relatively preserved.¹⁴ Other investigators have postulated that this redistribution of blood flow and oxygen delivery accounts for the numeric discrepancy between Svo₂ and Scvo₂.^{7,11,12,14,47,48} During states of circulatory collapse, Scvo₂ values are higher when compared with Svo₂, reflecting regional differences in perfusion and oxygen delivery.^43,47 Ruokonen et al⁴⁹ measured blood flow distribution and regional oxygen delivery in septic shock before and after vasopressor therapy. Oxygen delivery and oxygen consumption increased dramatically in splanchnic and leg blood flow during vasopressor therapy. This was out of proportion to the changes seen in systemic oxygen delivery, and thus the authors concluded that regional changes in oxygen delivery in septic shock cannot predict systemic changes. Sander et al⁵⁰ compared Svo₂ and Scvo₂ in patients with cardiac conditions and found that Scvo₂ overestimated Svo₂ when Svo₂ levels were low. Conversely, Scvo₂ underestimated Svo₂ when Svo₂ levels were high. The regional oxygen extraction rate was the principal difference between Svo₂ and Scvo₂. Similarly, differences in regional extraction may account for the lack of correlation of Scvo₂ and Sfvo₂ in our study.

There were several limitations to our study. First, the sample size was relatively modest, and few patients met the definition of shock. While some patients did meet criteria for early goal-directed therapy on admission, the venous samples were obtained at any point in a patient’s ICU stay, specifically when the femoral and nonfemoral catheters were both in place. Therefore, the samples were not necessarily drawn during a state of shock or prior to resuscitation. Few patients had an Scvo₂ in the range of interest (15% of the subjects had venous saturations of < 60%). Future studies should examine patients in shock prior to resuscitation. We also did not compare changes, but we recorded absolute values of Scvo₂ and Sfvo₂. Previous studies have found a lack of correlation between individual values of Scvo₂ and Svo₂, however changes in the two values were found to correlate.^19,24 A future study comparing changes before and after intervention may yield a better correlation. We also measured intermittent values of venous oxygen saturation rather than continuous values.

Rapid identification and intervention based on markers of tissue hypoperfusion and dysoxia is vital to the treatment of patients in shock.^{13,14,21-25,51,52} Given the challenges of inserting a subclavian or internal jugular intravenous catheter in the emergency setting and the frequency in which femoral catheters are placed as demonstrated by our survey, Sfvo₂ is an attractive substitute for Scvo₂ as a marker to guide resuscitative efforts. According to our study, Scvo₂ and Sfvo₂ did not have a consistent correlation, and the use of Sfvo₂ could lead to inappropriate interventions based on narrow resuscitation end points. These data need to be confirmed in a larger cohort of patients.

Author contributions:Dr Davison: contributed by obtaining study participants, writing, and revising the manuscript.

Dr Chawla: contributed by overseeing the research project, developing the original research protocol, performing all statistical analyses, and helping revise the manuscript.

Dr Selassie: contributed by developing the initial research protocol and obtaining study participants.

Ms Jones: contributed by obtaining study participants.

Ms McHone: contributed by obtaining study participants.

Ms Vota: contributed by obtaining study participants.

Dr Junker: contributed by helping revise the manuscript.

Dr Sateri: contributed to the development, administration, and collection of the survey data.

Dr Seneff: contributed by helping revise the manuscript.

Financial/nonfinancial disclosures: The authors have reported to CHEST the following conflicts of interest: Dr Junker consults for Edwards Lifesciences Corporation. Drs Davison, Chawla, Selassie, Sateri, and Seneff; and Mss Jones, McHone, and Vota have reported that no potential conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article.

Other contributions: We are grateful to the various physicians and nurses who made this study possible. We would like especially to thank Christina Seneff, Ermira Brasha-Mitchell, MD, Vikramjeet Saini, MD, and Margot Kern for their dedication and support in bringing this project to fruition.