Journal of Critical Care 28 (2013) 1111.e1–1111.e5
Contents lists available at ScienceDirect
Journal of Critical Care
j ourna l homepage: www. jcc journa l .org
Tissue oxygen saturation for the risk stratification of septic patients☆
Stefan W. Leichtle, MD a,⁎, Christodoulos Kaoutzanis, MD a, Mary-Margaret Brandt, MD a,Kathleen B. Welch, MS, MPH b, Mary-Anne Purtill, MD a
a Section of Surgical Critical Care, St Joseph Mercy Health System, Ann Arbor, MI 48106, USAb Center for Statistical Consultation and Research, University of Michigan, Ann Arbor, MI 48109, USA
a b s t r a c ta r t i c l e i n f o
☆ The authors declare that they have no conflicts of indisclosures.⁎ Corresponding author. Tel.: +1 734 846 1648; fax:
Tissue oxygen saturationNear-infrared spectroscopyStO2SepsisRapid response team
Purpose: Peripheral tissue oxygen saturation (StO2) has shown promise as an early indicator of tissuehypoperfusion and as a risk stratification tool in various forms of shock. The purpose of this study was todetermine if StO2 would predict admission to an intensive (ICU) or progressive care unit in patients with earlysigns of sepsis.Methods: In this prospective observational study, a rapid response team measured StO2 levels in patients
screening positive for sepsis. Using a logistic regression model, the value of StO2 as a predictor for ICUadmission within 72 hours of the initial assessment was determined.Results: The 31 (47%) of 66 patients who required ICU admission within 72 hours of evaluation had asignificantly lower StO2 value (median, 78% vs 81%; P= .05). All patients with StO2 less than 70% required ICUadmission. A 1-point increase in StO2 was associated with a 7% decrease in the odds of requiring ICUadmission, and the area under the curve for StO2 was 0.64 (0.51-0.77, P = .01).Conclusions: Low StO2 levels in patients screening positive for sepsis are associated with an increased risk ofICU admission, but their reliability as a predictor is rather low. An StO2 below 70% might be an interestingcutoff value for further study.
Sepsis is a major cause for morbidity and mortality, particularly inthe surgical population [1]. Early recognition and initiation oftreatment are critical to achieve satisfactory patient outcomes [2,3].Commonly determined hemodynamic parameters such as bloodpressure, heart rate, and oxygen saturation are often unreliable orlate indicators of increasing disease severity and do not correlate withthe microvascular perfusion abnormalities underlying sepsis [4,5].More advanced diagnostic parameters such as mixed venous oxygensaturation require invasive procedures, and lactic acid levels require aperipheral blood draw and results are not immediately available;moreover, these parameters may still not directly correlate with theextent and severity of tissue hypoperfusion. Early detection ofpatients at risk for more serious septic complications represents aparticular challenge on general medical and surgical wards, wheremonitoring capabilities are limited and continuous monitoring of vitalsigns, blood draws, or arterial blood gas determinations are notroutinely performed.
Based on the observation that tissue perfusion in developing shockdecreases first and recovers last in the periphery [6], the potential ofperipheral tissue oxygen saturation (StO2) to detect global hypoperfu-sion or classify disease severity has been studied in a variety ofsettings. Tissue oxygen saturation levels were shown to predictseverity of shock, multiorgan dysfunction, transfusion requirements,and death in patients with trauma [7-12] and to rapidly reflectcompromise of arterial blood supply [13,14]. Tissue oxygen saturationwas also found to correlate with advanced markers of cardiovascularstatus such as mixed central venous oxygen saturation [8] and wasstudied to classify disease severity and predict outcomes in septicpatients [15,16]. We hypothesized that StO2 measurements obtainedin patients on general medical and surgical wards who undergoevaluation by a rapid response team for suspected sepsis wouldidentify patients at risk for impending clinical deterioration.
2. Methods
All patients hospitalized on a general medical and surgical wardwho underwent evaluation by the rapid response team for a clinicalpresentation concerning for sepsis were assessed for inclusion in thestudy. Exclusion criteria were current hospitalization on an interme-diate or intensive care unit (ICU), established diagnosis of severesepsis or septic shock, age younger than 18 years, and currentpregnancy. At this institution, patients on a surgical andmedical wardwho demonstrate signs of systemic inflammatory response syndrome
The normally distributed variables temperature, heart rate, mean arterial pressure, andwhite blood cell count are expressed as mean ± SD; the nonnormally distributedvariables age, StO2, respiratory rate, pulse oximetry, and lactic acid are expressed asmedian (25th and 75th percentile).⁎ Denotes statistical significance (P ≤ .05).
1111.e2 S.W. Leichtle et al. / Journal of Critical Care 28 (2013) 1111.e1–1111.e5
undergo immediate evaluation by the rapid response team. Measure-ments taken by the rapid response team include determination oftemperature, heart rate, mean arterial pressure, respiratory rate, andoxygen saturation. In case of suspected or documented underlyinginfection, additional parameters includingwhite blood cell count, lacticacid level, and StO2 are determined. Included in this studywere patientswith systemic inflammatory response syndrome (defined as ≥2 of thefollowing: temperature N100.4F or b96.8F, heart rate N90 min−1,
respiratory rate N20 min−1, white blood cell count N12 000 mm−3 orb4000 mm−3) and a documented or suspected source of infection. Thestudy was approved by the institutional review board of Saint JosephMercy Health System, and a waiver of consent was granted.
Tissue oxygen saturation levels were obtained using near-infraredspectroscopy, which allows measurement of oxygenation levels intissue several millimeters underneath the skin surface [17]. Allmembers of the rapid response team were trained in the use of theInSpectra StO2 Tissue Oxygenation Monitor (Hutchinson TechnologyInc, Hutchinson, Minn). At bedside, the StO2 probe (model 1615;depth, 15 mm) was placed on the thenar eminence of the patient'shand, and the StO2 value displayed 20 seconds after the application ofthe probe was recorded. A member of the rapid response teamrecorded all obtained measurements as well as the date and time ofassessment in a separate data collection sheet for each patient, whichwas stored in a locked file cabinet until completion of the study.
After conclusion of the prospective data collection by the rapidresponse team, the data sheets were evaluated by 2 study in-vestigators (S.L. and C.K.). The clinical course of each patient in the 72hours after assessment by the rapid response teamwas then reviewedin the electronic medical records. The outcome of interest wasadmission to a monitored bed in the surgical or medical ICU orprogressive care unit during the 72 hours after assessment as a sign ofthe development of septic complications. This period was chosen tocapture not only patients who required immediate transfer to an ICUafter assessment but also those who deteriorated secondary to septiccomplications within the ensuing 3 days. The decision to transfer apatient was made by the physician of the patient's primary team, notthe rapid response team. Those patients admitted to the ICU weredesignated as “group 1,” and those patients whowere not admitted tothe ICU were designated as “group 2.”
All continuous demographic and clinical variables were initiallyassessed for normality using the Kolmogorov-Smirnoff test. Variablesthat were normally distributed, including temperature, heart rate,mean arterial pressure, and white blood cell count, were thensummarized using means and SDs, and a comparison of the meansof patients in groups 1 and 2wasmade using an independent-samplest test. Continuous variables that were not normally distributed, whichincluded age, StO2, respiratory rate, oxygen saturation, and lactic acid,were summarized using medians and 25th/75th percentiles. Thenonparametric Wilcoxon test was then used to compare the values ofthese variables between patients in groups 1 and 2. For the categoricalvariable “gender,” the proportion in each groupwas compared using aPearson χ2 test.
A simple logistic regression model was fit using each clinicalvariable individually to evaluate eachmeasure as a potential predictorof ICU admission. The Nagelkerke R-square and the area under thecurve were calculated for each measure. The median and the lowerand upper 95% confidence interval (CI) of the area under the curvewere calculated using 2000 bootstrap simulations for each measure.The Nagelkerke R-square is similar to the R-square in a linearregression model and ranges from 0 for a poor model to 1.0 for aperfect model. The Nagelkerke R-square gives an idea of theperformance of the predictors in a model. The area under the curveprovides an assessment of the use of a diagnostic test for distinguish-ing those patients who have a disease or outcome of interest vs thosewho do not. It can be interpreted as the probability that a randomlyselected person with the outcome of interest (in our study, being
transferred to the ICU) has a higher (or lower) value on the test than arandomly selected person who does not have the outcome of interest.To be a useful diagnostic tool, the 95% CI of the area under the curveshould be more than 0.5.
Finally, the area under the curve of each of the predictors wascompared using a logistic regression model to assess the relativeability of each of these clinical measures to predict ICU admissioncompared with StO2. Statistical analysis was performed using SAS 9.2,SAS Institute Inc, Cary, NC.
3. Results
Over a 1-year period, 66 data collection sheets were included inthe analysis. Thirty-one patients (47%) were transferred to an ICU orprogressive care unit within 72 hours of the initial assessmentbecause of clinical deterioration (group 1). All transfers had apreliminary diagnosis of sepsis and occurred after a median of lessthan 2 hours.
Their median StO2 was 78% (25th/75th percentiles, 70%/84%);temperature, 100.0F ± 1.8F; heart rate, 105 ± 19 min−1; meanarterial pressure, 82 ± 22 mm Hg; respiratory rate, 22 min−1 (20min−1/28 min−1); oxygen saturation, 96% (94%/99%); white bloodcell count, 16.8 ± 9.2 × 103/μL; and lactic acid, 2.1 IU/L ([1.3 IU/L]/[2.6 IU/L]). Thirty-five patients (53%) did not require ICU orprogressive care unit admission within 72 hours (group 2). Theirmean StO2 was 81% (77%/85%); temperature, 100.2F ± 1.9F; heartrate, 107 ± 20 min−1; mean arterial pressure, 83 ± 15 mm Hg;respiratory rate, 20min−1 (18min−1/22min−1); SpO2, 96 (92%/97%);white blood cell count, 15.3 ± 7.7 × 103/μL; and lactic acid, 1.6 IU/L([1.2 IU/L]/[2.2 IU/L]).
There were no statistically significant differences in age and sexbetween the patients in the 2 groups, but StO2 and respiratory ratewere significantly different between group 1 and group 2, with P =.05 and P = .05, respectively (Table 1). The distribution of StO2measurements in both groups is shown in Fig. 1. All patients with anStO2 less than 70% required admission to the ICU.
A simple logistic regression model including all 66 patientsdemonstrated a significant association between StO2 levels and ICUadmission, with an odds ratio of 0.93 (95% CI, 0.86-0.98; P = .03),indicating that a 1-point increase in StO2 was associated with a 7%decrease in the odds of requiring ICU admission. The area under thecurve for StO2, based on 56 patients with complete data, wascalculated to be 0.64 (95% CI, 0.51-0.77). The Nagelkerke pseudo–R-square for StO2 was 0.13. The other significant predictor for ICUadmission was respiratory rate, with an area under the curve of 0.65(95% CI, 0.51-0.77). Table 2 shows the area under the curve for allvariables used in the sepsis algorithm and their significance as a
Fig. 1. Distribution of StO2 measurements in patients who required ICU admission (bottom) and those who did not (top).
1111.e3S.W. Leichtle et al. / Journal of Critical Care 28 (2013) 1111.e1–1111.e5
predictor for ICU admission. The receiver operating characteristiccurve for StO2 is shown in Fig. 2.
4. Discussion
In this prospective observational study, lower StO2 levels deter-mined via spot-check measurement in patients with early signs ofsepsis were found to be associated with clinical deterioration, leadingto ICU admission within 72 hours of the initial assessment. However,StO2 was not found to be a reliable diagnostic tool for predicting ICUadmission, based on the relatively low area under the curve, and thelower confidence limit being very close to 0.50 (area under the curve,0.64; 95% CI, 0.51-0.77). Patients admitted to the ICU had significantlylower StO2 levels upon initial assessment, but the absolute differencewas small (median of 78% vs 81%, P= .05). Clinically, perhaps of moreinterest was the observation that all patients with an initial StO2 of lessthan 70% eventually required ICU admission, indicating that a spot-check StO2 less than 70%might be considered a strong predictor of ICUadmission. However, several patients with higher StO2 levels alsorequired ICU admission, implying that an StO2 above 70% does notreliably exclude this adverse outcome.
Table 2Area under the curve for all variables used in the sepsis algorithm and their significanceas predictor for admission to the ICU
Patients who required ICU admission also had a slightly yetsignificantly higher respiratory rate (22 min−1 vs 20 min−1), but thearea under the curve calculations for respiratory rate suggested anequally limited reliability as a diagnostic tool for predicting ICUadmission (95% CI, 0.51-0.77). None of the other obtained hemody-namic and laboratory parameters was found to be a significantpredictor of the adverse outcome. Of note, almost half of all patientsassessed for early signs of sepsis on the general surgical and medicalwards required admission to the ICU within 72 hours of theassessment, underscoring the importance of having an algorithm orprotocol for the early detection of sepsis.
Despite tremendous improvements in the care of septic patientsowing to the introduction of early goal–directed therapy [2] and the
Fig. 2. Receiver operating characteristic (ROC) curve of StO2 for predicting ICU admission.
1111.e4 S.W. Leichtle et al. / Journal of Critical Care 28 (2013) 1111.e1–1111.e5
efforts of the Surviving Sepsis Campaign [18], approximately 750 000patients hospitalized with severe sepsis create health care expendi-tures of more than $16 billion per year [19,20]. Mortality rates rangefrom 20% to more than 50% and are inversely related to earlyrecognition and initiation of treatment [21-23]. Several studiesperformed in a variety of settings with patients in the state of shockhave suggested that StO2 levels might be an early marker of globalhypoperfusion and a predictor of patient outcomes [7-11,13-16].
In our study, there was only a 3-point difference in median StO2levels between patients with sepsis eventually requiring ICUadmission and those who did not. This difference was statisticallysignificant but may be of limited use in clinical practice. Thelimitations of early StO2 measurements in discriminating betweenpatients who will do well and those who will not were also shown byVorwerk and Coats [24], who found that thenar StO2 levels of patientswith severe sepsis did not significantly differ between survivors (StO2,72%) and nonsurvivors (StO2, 72%) upon arrival in the emergencydepartment. Tissue oxygen saturation levels later in the course ofdisease were studied by Payen et al [25], who demonstrated asignificant difference in baseline StO2 of patients with septic shock(82%; 75%-88%) and healthy volunteers (89%; 85%-92%). Mulier et al[26] also demonstrated significantly lower StO2 levels in patients withsevere sepsis (75%-83%) than in healthy volunteers. Beekley et al [12]demonstrated that minimum StO2 levels were an early predictor ofblood transfusions in combat casualties. These findings suggest thatStO2 levels may be unreliable for the early identification of sepsis, butmight be more useful in identifying global hypoperfusion in otherforms of shock and in later stages of sepsis.
We also found a lack of correlation between StO2 levels and othercommonly obtained hemodynamic and laboratory parameters inseptic patients, with the exception of a marginally significantnegative correlation between StO2 levels and lactic acid (r=−0.24,P = .05). Lima et al [27] studied StO2 levels in 22 critically illpatients undergoing early goal–directed therapy and did not find asignificant relationship between StO2, heart rate, and mean arterialpressure. Similar observations were made for lactic acid [26], mixedvenous oxygen saturation [28], mixed central venous oxygensaturation [27], base deficit [26], and Acute Physiology and ChronicHealth Evaluation II scores [26] in other studies. Podbregar andMozina [28] demonstrated a lack of correlation between StO2 andSvO2 measurements in patients with severe sepsis and heart failure,whereas there was a good correlation in patients with severe leftheart failure alone and in healthy volunteers, suggesting aninfluence of cardiac function in this situation.
An alternative to single, spot-check measurements of StO2 levelsmight be dynamic StO2 measurements, as performed in studies byPayen et al [25] and Skarda et al [29]. Tissue oxygen saturationreperfusion slopes after vascular occlusion tests reliably differentiatedpatients with severe sepsis from health volunteers [29] and correlatedwell with hemodynamic and metabolic parameters in patients withseptic shock [25]. In addition, StO2 levels might have to bemeasured atseveral time points; Lima et al [27] measured StO2 levels every 2 hoursin patients undergoing resuscitation and found that StO2 levels below70% at 8 hours after resuscitation correlatedwith increased SequentialOrgan Failure Assessment and Acute Physiology and Chronic HealthEvaluation II scores.
Vorwerk and Coats also followed multiple StO2 measurements inpatients undergoing resuscitation and found a significant improve-ment from baseline StO2 levels of 72% to 78% in survivors but not innonsurvivors. Overall, patients who persisted to have StO2 levelsbelow 75% after resuscitation efforts had a 2-fold increased risk ofmortality. These StO2 values demonstrated by Lima et al and Vorwerkand Coats are close to the 70% mark, below which all patients in ourstudy required ICU admission, suggesting that an StO2 less than 70%might be of practical use to identify a patient group at high risk forcomplications and adverse outcomes.
It is difficult to determine a range of truly normal StO2 levels.Crookes et al [10] compared StO2 levels in a control group with thosein patients with trauma and reported StO2 levels of 87%% ± 6% for thecontrol group, 83%% ± 10% in patients with trauma without shock,83%% ± 10% in patients with mild shock, 80% ± 12% in patients withmoderate shock, and 45% ± 26% in patients with severe shock. Mulieret al [26] reported StO2 levels in patients with severe sepsis to be 75%±15%, 76% ± 17%, and 83% ± 7% on days 1, 2, and 3 of the assessment,respectively, and Podbregar and Mozina [28] demonstrated thenarStO2 levels to be 84% ± 4% in healthy volunteers, 58% ± 13% inpatients with heart failure, and 90%± 7% in patients with heart failureand septic shock. Tissue oxygen saturation levels in septic patientsappear to range from 72% [15] to 90% [28], which includes thecommonly assumed normal values. In septic patients, StO2 levels maybe either low as a sign of global hypoperfusion (in septic shock) or high,secondary to inadequate oxygen use. Moreover, the results of a studyby Soller et al [30] suggested that the body site used for themeasurement of StO2 levels might be important. In 16 healthyvolunteers exposed to progressively lower body negative pressure, itwas shown that spectroscopically determined forearm muscle oxygensaturation progressively decreased with decreasing stroke volume andblood pressure, whereas StO2 levels recorded at the thenar eminencedid not change significantly throughout the study course.
This study is limited in that our findings stem from 66 patients, anda more robust capability of StO2 to predict ICU admission mighttherefore not have been detected secondary to this relatively smallnumber of patients. The single spot-check StO2 measurement after 20seconds of calibration obtained by the rapid response team mighthave been inadequate to truly reflect the changes in macroperfusionand microperfusion in a complex disease process such as sepsis.Because the study was designed to be pragmatic and to avoid addingsubstantial work for the rapid response team, the StO2 measurementswere not optimized, for example, by performing an ischemicchallenge with a vascular occlusion test. Moreover, StO2 levels inthis study were obtained at a very early stage of sepsis, as reflected bythe lack of differences in most other hemodynamic and laboratoryvalues between the 2 patient groups. The retrospective review of theelectronic medical records to determine if and when patients weretransferred to an ICU or progressive care unit was sometimes limitedby unclear or inaccurate documentation; therefore, it was frequentlynot possible to determine an exact diagnosis or etiology of severesepsis at the time of transfer.
In summary, spot-check StO2 measurements in patients with earlysigns of sepsis were significantly lower in patients who required ICUadmission within 72 hours of the initial assessment. In addition,because all patients with StO2 less than 70% required ICU admission,this cutoff value might be of interest for further study. The smallabsolute difference in StO2 and the area under the curve calculationsfor StO2 demonstrated that the overall clinical reliability as a tool forearly risk stratification might be limited.
References
[1] Moore LJ, Moore FA, Jones SL, et al. Sepsis in general surgery: a deadlycomplication. Am J Surg 2009;198(6):868–74.
[2] Rivers E, Nguyen B, Havstad S, et al. Early goal-directed therapy in the treatment ofsevere sepsis and septic shock. N Engl J Med 2001;345(19):1368–77.
[3] Moore LJ, Jones SL, Kreiner LA, et al. Validation of a screening tool for the earlyidentification of sepsis. J Trauma 2009;66(6):1539–46 [discussion 1546–1537].
[4] Spronk PE, Zandstra DF, Ince C. Bench-to-bedside review: sepsis is a disease of themicrocirculation. Crit Care 2004;8(6):462–8.
[5] Trzeciak S, Dellinger RP, Parrillo JE, et al. Early microcirculatory perfusionderangements in patients with severe sepsis and septic shock: relationship tohemodynamics, oxygen transport, and survival. Ann Emerg Med. 2007; 49(1): 88–98, 98 e81-82.
[6] Poeze M, Solberg BC, Greve JW, et al. Monitoring global volume-relatedhemodynamic or regional variables after initial resuscitation: what is a betterpredictor of outcome in critically ill septic patients? Crit Care Med 2005;33(11):2494–500.
1111.e5S.W. Leichtle et al. / Journal of Critical Care 28 (2013) 1111.e1–1111.e5
[7] Cohn SM, Nathens AB, Moore FA, et al. Tissue oxygen saturation predicts thedevelopment of organ dysfunction during traumatic shock resuscitation. J Trauma2007;62(1):44–54 [discussion 54–45].
[8] Mesquida J, Masip J, Gili G, et al. Thenar oxygen saturation measured by nearinfrared spectroscopy as a noninvasive predictor of low central venous oxygensaturation in septic patients. Intensive Care Med 2009;35(6):1106–9.
[9] Moore FA, Nelson T, McKinley BA, et al. Massive transfusion in trauma patients:tissue hemoglobin oxygen saturation predicts poor outcome. J Trauma2008;64(4):1010–23.
[10] CrookesBA,CohnSM,BlochS, et al. Cannear-infrared spectroscopy identify the severityof shock in trauma patients?J Trauma 2005;58(4):806–13 [discussion 813–806].
[11] McKinley BA, Marvin RG, Cocanour CS, et al. Tissue hemoglobin O2 saturationduring resuscitation of traumatic shock monitored using near infrared spectrom-etry. J Trauma 2000;48(4):637–42.
[12] Beekley AC, Martin MJ, Nelson T, et al. Continuous noninvasive tissue oximetry inthe early evaluation of the combat casualty: a prospective study. J Trauma2010;69(Suppl 1):S14–25.
[13] Gomez H, Torres A, Polanco P, et al. Use of non-invasive NIRS during a vascularocclusion test to assess dynamic tissue O(2) saturation response. Intensive CareMed 2008;34(9):1600–7.
[14] Putnam B, Bricker S, Fedorka P, et al. The correlation of near-infrared spectroscopywith changes in oxygen delivery in a controlled model of altered perfusion. AmSurg 2007;73(10):1017–22.
[15] Creteur J, Carollo T, Soldati G, et al. The prognostic value of muscle StO2 in septicpatients. Intensive Care Med 2007;33(9):1549–56.
[16] Leone M, Blidi S, Antonini F, et al. Oxygen tissue saturation is lower innonsurvivors than in survivors after early resuscitation of septic shock.Anesthesiology 2009;111(2):366–71.
[17] Lipcsey M, Woinarski NC, Bellomo R. Near infrared spectroscopy (NIRS) of thethenar eminence in anesthesia and intensive care. Ann Intensive Care2012;2(1):11.
[18] Dellinger RP, Levy MM, Carlet JM, et al. Surviving Sepsis Campaign: internationalguidelines for management of severe sepsis and septic shock: 2008. Crit Care Med2008;36(1):296–327.
[19] Angus DC, Linde-Zwirble WT, Lidicker J, et al. Epidemiology of severe sepsis in theUnited States: analysis of incidence, outcome, and associated costs of care. CritCare Med 2001;29(7):1303–10.
[20] Martin GS, Mannino DM, Eaton S, et al. The epidemiology of sepsis in the UnitedStates from 1979 through 2000. N Engl J Med 2003;348(16):1546–54.
[21] Gaieski DF, Mikkelsen ME, Band RA, et al. Impact of time to antibiotics on survivalin patients with severe sepsis or septic shock in whom early goal–directed therapywas initiated in the emergency department. Crit Care Med 2010;38(4):1045–53.
[22] Kumar A, Roberts D, Wood KE, et al. Duration of hypotension before initiation ofeffective antimicrobial therapy is the critical determinant of survival in humanseptic shock. Crit Care Med 2006;34(6):1589–96.
[23] Jones AE, Brown MD, Trzeciak S, et al. The effect of a quantitative resuscitationstrategy on mortality in patients with sepsis: a meta-analysis. Crit Care Med2008;36(10):2734–9.
[24] Vorwerk C, Coats TJ. The prognostic value of tissue oxygen saturation inemergency department patients with severe sepsis or septic shock. Emerg MedJ 2012;29(9):699–703.
[25] Payen D, Luengo C, Heyer L, et al. Is thenar tissue hemoglobin oxygen saturation inseptic shock related to macrohemodynamic variables and outcome? Crit Care2009;13(Suppl 5):S6.
[26] Mulier KE, Skarda DE, Taylor JH, et al. Near-infrared spectroscopy in patients withsevere sepsis: correlation with invasive hemodynamic measurements. Surg Infect(Larchmt) 2008;9(5):515–9.
[27] Lima A, van Bommel J, Jansen TC, et al. Low tissue oxygen saturation at the end ofearly goal–directed therapy is associated with worse outcome in critically illpatients. Crit Care 2009;13(Suppl 5):S13.
[28] Podbregar M, Mozina H. Skeletal muscle oxygen saturation does not estimatemixed venous oxygen saturation in patients with severe left heart failure andadditional severe sepsis or septic shock. Crit Care 2007;11(1):R6.
[29] Skarda DE, Mulier KE, Myers DE, et al. Dynamic near-infrared spectroscopymeasurements in patients with severe sepsis. Shock 2007;27(4):348–53.
[30] Soller BR, Ryan KL, Rickards CA, et al. Oxygen saturation determined from deepmuscle, not thenar tissue, is an early indicator of central hypovolemia in humans.Crit Care Med 2008;36(1):176–82.
