Share this article on:

Trends of Hemoglobin Oximetry: Do They Help Predict Blood Transfusion During Trauma Patient Resuscitation?

Yang, Shiming PhD*; Hu, Peter F. PhD*; Anazodo, Amechi MD, MPH; Gao, Cheng MS; Chen, Hegang PhD; Wade, Christine MS; Hartsky, Lauren MA§; Miller, Catriona PhD§; Imle, Cristina MS; Fang, Raymond MD§; Mackenzie, Colin F. MBChB*

doi: 10.1213/ANE.0000000000000927
Technology, Computing, and Simulation: Research Report

BACKGROUND: A noninvasive decision support tool for emergency transfusion would benefit triage and resuscitation. We tested whether 15 minutes of continuous pulse oximetry–derived hemoglobin measurements (SpHb) predict emergency blood transfusion better than conventional oximetry, vital signs, and invasive point-of-admission (POA) laboratory testing. We hypothesized that the trends in noninvasive SpHb features monitored for 15 minutes predict emergency transfusion better than pulse oximetry, shock index (SI = heart rate/systolic blood pressure), or routine POA laboratory measures.

METHODS: We enrolled direct trauma patient admissions ≥18 years with prehospital SI ≥0.62, collected vital signs (continuous SpHb and conventional pulse oximetry, heart rate, and blood pressure) for 15 minutes after admission, and recorded transfusion (packed red blood cells [pRBCs]) within 1 to 3, 1 to 6, and 1 to 12 hours of admission. One blood sample was drawn during the first 15 minutes. The laboratory Hb was compared with its corresponding SpHb reading for numerical, clinical, and prediction difference. Ten prediction models for transfusion, including combinations of prehospital vital signs, SpHb, conventional oximetry, and routine POA, were selected by stepwise logistic regression. Predictions were compared via area under the receiver operating characteristic curve by the DeLong method.

RESULTS: A total of 677 trauma patients were enrolled in the study. The prediction performance of the models, including POA laboratory values and SI (and the need for blood pressure), was better than those without POA values or SI. In predicting pRBC 1- to 3-hour transfusion, adding SpHb features (receiver operating characteristic curve [ROC] = 0.65; 95% confidence interval [CI], 0.53–0.77) does not improve ROC from the base model (ROC = 0.64; 95% CI, 0.52–0.76) with P = 0.48. Adding POA laboratory Hb features (ROC = 0.72; 95% CI, 0.60–0.84) also does not improve prediction performance (P = 0.18). Other POA laboratory testing predicted emergency blood use with ROC of 0.88 (95% CI, 0.81–0.96), significantly better than the use of SpHb (P = 0.00084) and laboratory Hb (P = 0.0068).

CONCLUSIONS: SpHb added no benefit over conventional oximetry to predict urgent pRBC transfusion for trauma patients. Both models containing POA laboratory test features performed better at predicting pRBC use than prehospital SI, the current best noninvasive vital signs transfusion predictor.

Supplemental Digital Content is available in the text.Published ahead of print August 21, 2015

From the *Department of Anesthesiology, University of Maryland School of Medicine, Baltimore Maryland; Program in Trauma, R. Adams Cowley Shock Trauma Center, University of Maryland School of Medicine, Baltimore, Maryland; Department of Epidemiology, University of Maryland School of Medicine, Baltimore, Maryland; §U.S. Air Force Center for the Sustainment of Trauma and Readiness Skills, University of Maryland, Baltimore, Maryland.

Accepted for publication June 16, 2015.

Published ahead of print August 21, 2015

Funding: This research was funded by U.S. Air Force (FA8650-11-2-6D01) Continuing Noninvasive Monitoring and the Development of Predictive Triage Indices for Outcomes Following Trauma. Masimo (Masimo Corporation, Irvine, CA) provided the SpHb monitors but had no role in the design, execution, or analysis of this research.

The authors declare no conflicts of interest.

Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s website.

Address correspondence to Colin F. Mackenzie, MBChB, Departments of Anesthesiology and Physiology, University of Maryland School of Medicine, 11 S Paca St., Suite LL01, Baltimore, MD 21201. Address e-mail to cmack003@umaryland.edu.

Hemorrhage is the most common cause of preventable death on the battlefield and in civilian trauma care.1,2 The ability to distinguish rapidly and accurately between those patients with and those without life-threatening bleeding in the field and during point-of-admission (POA) care is a key and, as yet imperfectly realized, the element of trauma triage for both the initiation of primary control of bleeding and the timely provision of the appropriate range of blood products.

Diagnosis of bleeding is difficult when the location of hemorrhage is not obvious, such as in the chest, head, or abdomen, and when the extent of hemorrhage is unknown. Therefore, robust and quickly identifiable evidence would be useful for guiding early imaging, hemorrhage-control, and life-saving interventions. Hemoglobin (Hb) concentration is an important consideration in patient assessment during hemorrhage and trauma care, although it has limitations.3 Invasive laboratory Hb measurement provides a single reading at the sampling time. It is subject to dilutional variation, and results take time to be processed. Pulse oximetry is an inexpensive, simple, and noninvasive technology for patient monitoring.4,5 The Masimo Rainbow® Pulse CO-Oximetry™ (Masimo Corporation, Irvine, CA) estimates total Hb concentration and is used for noninvasive continuous hemoglobin (SpHb) monitoring.a Therefore, it has the potential for use in computer-aided algorithms, essentially immediately, to detect changes in Hb concentration status in trauma patients and to provide evidence to support clinicians’ early decision making and transfusion planning.

Our study evaluated the difference between SpHb and conventional laboratory Hb in an unstable trauma patient population during resuscitation. We also tested whether continuously measured SpHb at the point of trauma center admission can improve transfusion and mortality prediction. We hypothesized that the changing trends of SpHb features predict emergency transfusion better than prehospital shock index (SI = heart rate [HR]/systolic blood pressure; bpm/mm Hg), pulse oximetry alone or Hb alone, or routine POA laboratory measures, excluding Hb.

Back to Top | Article Outline

METHODS

With approval of a waiver of patients’ informed consent from the University of Maryland School of Medicine and U.S. Air Force IRBs, adult patients (aged ≥18 years) with abnormal prehospital SI (≥0.62) were consecutively enrolled when they were directly admitted to the Baltimore R Adams Cowley Shock Trauma Center, from December 2011 to May 2013. Enrollment occurred 24 hours per day and 7 days per week when patients met entry criteria and SpHb sensors could be placed without interruption of patient care. Any patient who later developed into an SI ≥0.62 was not eligible for enrollment. The Shock Trauma Center is the primary adult resource center for the State of Maryland and admits >5000 trauma patients per year. Continuous vital signs were collected via BedMaster (GE Marquette, Milwaukee, WI) vital signs collection system during the first hour of patient resuscitation. Before the study, we conducted a power analysis and estimated that the sample size for our statistical comparison would need to be at least 443 (Supplemental Digital Content 1, http://links.lww.com/AA/B199). As the consort diagram in Figure 1 shows, 1191 patients were admitted to the trauma resuscitation unit (TRU), satisfying the age and prehospital SI criteria. All patients were directly transported from the scene of injury; none received blood before hospital admission. After excluding 480 eligible patients in whom the placement of an additional sensor for the purposes of continuous SpHb monitoring did not occur for logistical reasons (see Study Limitations), 711 patients had continuous SpHb measurement. Within this subgroup, we removed 34 patients who had incomplete laboratory POA data. For outcomes, the use of blood was documented for intervals of 3, 6, and 12 hours after admission and validated via blood bank records.

Figure 1

Figure 1

Two approaches to obtain Hb values were used. After a patient’s admission to the TRU, a Masimo Rainbow Pulse CO-Oximetry with SpHb was applied (in addition to a conventional pulse oximeter and vital signs sensors) to allow continuous monitoring of SpHb. Masimo Rad-87 (ver. 1405) software was used. The SpHb sensor (Rev F) was usually placed on the finger on the opposite side to the blood pressure cuff. A black finger shield was secured over the finger sensor to prevent ambient light interference with the SpHb sensor, and the finger sensor was placed with one digit separation from the GE Marquette pulse oximeter used for patient care. If there were bilateral upper arm injuries, the sensor was applied to the left big toe.

A blood sample was drawn in tandem with IV access within the first 15 minutes after TRU admission. A colocated research assistant recorded the SpHb reading at the time of the laboratory draw. Blood samples were analyzed for Hb concentration (Sysmex® XN-2000 Automated Hematology Analyzer; Sysmex Corp., Kobe, Japan)b and other tests including partial thromboplastin time, international normalized ratio, fibrinogen, lactate, and glucose (Analyzer NSN 6630015205212; Abbott Laboratories Inc., Chicago, IL) within 5 to 15 minutes after sampling.

Back to Top | Article Outline

Statistical Analysis

Statistical analyses compared (1) absolute values of SpHb and laboratory Hb, (2) their transfusion prediction performance, and (3) the change in transfusion prediction with additional blood sample analyses. First, the difference between SpHb and laboratory Hb readings was directly compared using the Bland-Altman plot.6–8 Mean bias and 95% limits of agreement were used to quantify the difference between the 2 measurements.9 Because the Bland-Altman plot is an overall evaluation that does not distinguish subsets of different clinical meaning, a Clarke-type error grid analysis10 was used as another way to display the difference between SpHb, as a new measuring tool, and laboratory Hb, as the reference measurement, and to suggest the clinical relevance of the differences.11 Within the Clarke-type error plot, a linear regression and the coefficient of determination (R2) are shown to evaluate how well SpHb can be replicated by the linear model using laboratory Hb as the predictor.

Second, we compared the prediction performance of features derived from SpHb and laboratory Hb to show the usefulness of each measurement, even though they may be different. We asked whether noninvasive continuous SpHb monitoring could be as good as, or could improve, the prediction accuracy of invasive Hb measurement for mortality or for institution of life-saving interventions, such as blood transfusion. Third, we compared the mortality prediction performance with models using laboratory tests other than Hb concentration, including partial thromboplastin time, international normalized ratio, fibrinogen, lactate, and glucose, which are predictors of hemorrhage shock and mortality. Later, we described the design of predictive features and models, as well as their evaluations.

Back to Top | Article Outline

Vital Signs Features

We compared the usefulness (prediction power) of SpHb and Hb for dichotomous outcomes such as packed red blood cell (pRBC) use in the 1 to 3, 1 to 6, and 1 to 12 hours after admission and mortality. We included 4 types of features in the models and adjusted for age and sex. Because SI is used as the current best predictor of transfusion12,13 and HR is an important factor in circulatory assessment, prehospital HR and prehospital SI were included in the base models. For continuous SpHb measurements from the Masimo sensor, we used the first 15 minutes of data after the sensor placement, which was usually within 1 to 5 minutes of patient arrival. We designed various features to quantify the SpHb, including changing trend and “dose” below clinical thresholds, which are detailed later. Models also included other values analyzed from laboratory blood tests, obtained within the first 15 minutes of TRU admission that are available from 3 point-of-care testing cartridges (iSTAT; Abbott Laboratories Inc., Chicago, IL) related to detection of hemorrhage and shock states. Table 1 summarizes the 10 models in terms of their variables, which can be categorized into 2 groups. The first 5 models use the prehospital HR, whereas the last 5 models use the prehospital SI. Within each group, we compared models using the addition of SpHb, Hb, other laboratory tests, and all available information. We included only the main effects in the models, for parsimonious and robust model fitting (see Supplemental Digital Content 2, http://links.lww.com/AA/B200, for the significance tests of interaction terms in the models).

Table 1

Table 1

Figure 2

Figure 2

Thirty-one features were designed for continuous SpHb analysis and used for selection in prediction models. The degree and duration of SpHb <10, 11, 12, 13, and 14 g/dL were calculated.14 The first, second, and third quartiles of SpHb, as well as its interquartile range and changing trend of SpHb, were assessed as features of the models.7 To characterize the changing pattern of SpHb during the initial 15-minute monitoring, SpHb values were averaged within a sequence of exclusive same size time windows, that is, 1, 2, and 3 minutes (red, black, and green curves in Fig. 2). The rate of change was calculated between any 2 averaged SpHb values in the same size window. For example, when the first 15 minutes of continuous SpHb measurements were averaged every minute, there were 15 data points (denoted as Avei, 1≤ i ≤15). The slope between 2 distinct data points i and j was calculated as Slopeij = (Avej − Avei)/(ji), with unit g/dL/min, where 1≤ i < j ≤15. We then calculated the percentage occurrence of increase and decrease, the maximum increase and decrease rate, and their standard deviations (SDs).

Back to Top | Article Outline

Models

Using multivariate logistic regression models, we compared feature groups for transfusion prediction performance in terms of the area under the receiver operating characteristic curve (AUROC). To avoid overfitting, we used stepwise feature selection to build parsimonious models. In forward selection, features with the level of significance of the Wald χ2 test ≤0.2 were included; in backward selection, features with the significance level >0.3 were removed. Furthermore, we used 10-fold cross-validation repeated 10 times with stratified sampling to examine how well-trained models could predict with previously unseen data. Because relatively few of the patients were transfused, the data were skewed, so the AUROC curve was used to evaluate the transfusion discriminant capability of each classification model,15 and the ROCs were compared by the DeLong method, with the null hypothesis that a pair of ROCs is not significantly different.16P value <0.05 was considered statistically significant. The Hosmer-Lemeshow goodness-of-fit test was implemented for checking the model fitting.17 All models’ prediction performances are reported with their ROCs, 95% confidence interval (CI) of ROCs, sensitivity, and specificity. All statistical analyses, predictive model building, and evaluation were implemented with R software version 3.1.1 (R Development Core Team, Vienna, Austria). Stepwise logistic regression used SAS 9.3 PROC LOGISTIC (SAS Institute Inc., Cary, NC).

Back to Top | Article Outline

RESULTS

Table 2 summarizes the demographics of the final data set in comparisons with the original 1191 cases. We enrolled 677 patients (479 males) with both continuous SpHb monitoring and POA laboratory tests, which included Hb.

Table 2

Table 2

The Bland-Altman analysis shows that the mean difference (bias) of the 2 measurements was −1.0 g/dL. The 95% limits of agreement ranged from 3.0 to −4.3 g/dL (bias ± 1.96 SD of the differences; see Fig. 3). SpHb readings were generally lower than Hb values. The histogram shows a normal distribution of differences centering near the bias; 35.2% of data points have differences within −1.0 to 1.0 g/dL and 64.8% have differences in the range −2.0 to 2.0 g/dL.

Figure 3

Figure 3

Figure 4 shows the modified Clarke-type error grid with a fitted linear line. The scatter plot of SpHb against laboratory Hb is partitioned into 3 regions with clinical meaning. With the Clarke-type error grid analysis, there are 98.93% points that fall into region A. In this data set, no data points fall into region C. Figure 5 shows that the histograms of SpHb (green) and laboratory Hb (red) were mostly located >10 g/dL. The laboratory Hb has a mean of 14.0 g/dL with an SD of 1.6 g/dL (first, second, and third quartiles are 13.0, 14.1, and 15.0 g/dL, respectively). The SpHb has a mean of 12.9 g/dL with an SD of 1.7 g/dL (first, second, and third quartiles are 11.8, 13.0, and 14.0 g/dL, respectively). Moreover, using SpHb as a predictor and laboratory Hb as a response, a linear regression model was fitted with the data. The coefficient of determination, R2 = 0.645, showed that the data fit only fairly with the linear equation, indicating that SpHb and laboratory Hb in this data set do not have a strong linear relationship.

Figure 4

Figure 4

Figure 5

Figure 5

At the level of predictive power comparison, all models had balanced training and testing performance, meaning that their AUROC differences were <10%. Therefore, we used the models’ training over the entire data set for profiling their prediction performance. Moreover, the Hosmer-Lemeshow goodness-of-fit test shows that there was no strong evidence of poor fit for each transfusion prediction model (see Supplemental Digital Content 1, http://links.lww.com/AA/B199). Tables 3, 4, 5, and 6 summarize the performance of models in predicting pRBC use in 3, 6, 12 hours, and mortality, measured by ROC, 95% CI of ROC, sensitivity, and specificity. In general, models with prehospital SI as a candidate feature had higher AUROCs than those with prehospital HR alone. In models that use either prehospital SI or HR, the prediction performance showed statistically significant differences among the feature groups of SpHb, laboratory Hb, and other laboratory tests. Using features from SpHb measured in the first 15 minutes does not significantly improve the prediction sensitivity and specificity from the base models. Although models using laboratory Hb have higher AUROC than models using SpHb (see Figs. 6 and 7), they both only have fair (AUROCs <0.8) performance.

Table 3

Table 3

Table 4

Table 4

Table 5

Table 5

Table 6

Table 6

Figure 6

Figure 6

Figure 7

Figure 7

Table 7

Table 7

Table 8

Table 8

Table 9

Table 9

Table 10

Table 10

However, other laboratory test results boost the model performance, especially in predicting blood transfusion. For example, the model using prehospital HR and other laboratory tests to predict pRBC 1 to 3 hours has AUROC = 0.88 (95% CI, 0.81–0.96), which is significantly higher than the model using SpHb (AUROC = 0.65; 95% CI, 0.53–0.77; P = 0.00084) and the model using laboratory Hb (AUROC = 0.72; 95% CI, 0.60–0.84; P = 0.00678) in predicting pRBC use in 1 to 3 hours after admission. Models using prehospital SI further support the inference that SpHb and laboratory Hb do not contribute significantly to transfusion prediction. By adding SpHb or Hb features to the models using laboratory tests, the performance showed no statistically significant difference. Also, the 95% CI of their ROCs are highly overlapped (Tables 3–6). ROC comparison through the DeLong methods shows that their ROCs are not significantly different (Tables 7–10). Hence, we conclude that SpHb and the Hb value in POA laboratory tests have no significant contribution to transfusion prediction, independent of the partial thromboplastin time, international normalized ratio, fibrinogen, lactate, and glucose. Tables 7 to 10 include the P values for model AUROC comparisons.

Back to Top | Article Outline

DISCUSSION

SpHb monitoring is an appealing new technique for estimating changes in Hb concentration. No studies have monitored transfusion prediction performance of SpHb trends or compared these with Hb in real time during trauma patient resuscitation. Several studies have compared SpHb with laboratory Hb with different populations monitored in stabilized conditions, most without SpHb trends. Studies compared SpHb and Hb during spine surgery,6,8 gastrointestinal bleeding in ICU,18 cardiac surgery,19 general ICU patients,20 and trauma patients in ICU.21 Trends in differences between SpHb measurements were monitored in one article studying ICU patients.20 In these articles, different statistical methods were used to quantify SpHb and Hb measurement agreement. Because of the experimental and patient population differences, conclusions on SpHb accuracy also vary. In a study on 20 patients undergoing spine surgery, SpHb and laboratory Hb had a difference of <1.5 g/dL for 61% of observations and a difference of >2.0 g/dL for 22% of observations.6 In 44 patients with repeated measures, a total of 85 pairs of SpHb and laboratory Hb analyzed by HemoCue, SpHb gave lower readings during surgery with bleeding,22 based on linear regression and Bland–Altman analysis. Another study using 165 laboratory Hb measurements obtained from 20 subjects undergoing hemodilution demonstrated an average difference of <1.0 g/dL compared with the laboratory Hb measurements.23 A meta-analysis of 32 studies also suggests that the wide limits of agreement between SpHb and laboratory Hb (−2.59 to 2.80 g/dL) should make clinicians cautious when using the SpHb values.24 Even 2 different noninvasive SpHb sensors (the Pronto-7 monitor [Masimo Corporation] and the NBM-200MP monitor [OrSense, Nes Ziona, Israel]) were found to have limited agreement.25

Despite the active research in quantifying the accuracy of SpHb compared with laboratory Hb, guidance in interpreting the difference is rare. Naftalovich and Naftalovich26 attempted to distinguish the macro-Hb measured by laboratory test and the total Hb estimated by SpHb using both macro- and micro-Hb. They hypothesized that more contribution from the microcirculatory Hb during blood loss increases the difference. In our study, through multivariate logistic regression and ROC evaluation, we demonstrated that the changing trend of SpHb does not boost the predictive models in comparison with base models that use only age, sex, and prehospital vital signs. The laboratory Hb values also did not significantly improve transfusion predictions from the base models. However, other laboratory tests, such as partial thromboplastin time, international normalized ratio, lactate, and glucose, significantly improve the discriminant capability of the transfusion predictive models, as these laboratory values are well recognized to changes with mediator release and reperfusion injury associated with mortality and accompanying resuscitation from hemorrhagic shock.27–29 A previous study also showed that SpHb may not improve the prediction of pRBC use compared with the features extracted from conventional pulse oximetry.30

In this single-center study, the following were limitations. First, not all patients admitted during the study enrollment period were included because of logistical reasons, such as multiple simultaneous admissions, concerns that adding an additional pulse oximeter and finger shield would interrupt emergency clinical care (especially in those mortally injured), and emergency patient admissions occurring with insufficient notice to set up the data collection process. Thus, the enrolled cohort had a greater incidence of blunt trauma, lower mortality, and transfusion rate than the entire patient admission cohort; therefore, SpHb may produce a different prediction of transfusion needs after penetrating injury, in military trauma and among very severely injured populations. Second, each patient had only one blood sample in the first 15 minutes after TRU admission. For accurate comparison and improved transfusion prediction power, repeated samples may reduce the impact of fluid bolus given during resuscitation and identify interval changes associated with bleeding.31 In addition, Masimo Corporation has revised the sensor used since study enrollment was completed and the new Rev K sensor may give different results. Because of proprietary issues, photoplethysmography data from Masimo pulse oximetry were not made available, so that perfusion index and pleth variability index could not be used in our prediction models. Finally, the outcomes of our data set are imbalanced (e.g., 3.0% patients were given pRBC in the 1–3 hours after admission), and this may cause biases affecting the accuracy of models for early transfusion prediction.

Back to Top | Article Outline

CONCLUSIONS

Noninvasive SpHb provides continuous monitoring limits of agreement with laboratory Hb that are too wide for clinical use during trauma patient resuscitation. Adding SpHb trend features, such as dose of changes in SpHb, rate of change, or identifying thresholds or changes of SpHb during the first 15 minutes of continuous measurement, may not improve the prediction of urgent pRBC transfusion for trauma patients in comparison with the use of base models, including prehospital HR or SI alone.

Back to Top | Article Outline

DISCLOSURES

Name: Shiming Yang, PhD.

Contribution: This author helped design the study, analyze the data, and write the manuscript.

Attestation: This author has seen the original study data, reviewed the analysis of the data, approved the final manuscript, and is the author responsible for archiving the study files.

Name: Peter F. Hu, PhD.

Contribution: This author helped design the study, analyze the data, and write the manuscript.

Attestation: This author has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.

Name: Amechi Anazodo, MD, MPH.

Contribution: This author helped design the study and conduct the study.

Attestation: This author has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.

Name: Cheng Gao, MS.

Contribution: This author helped analyze the data.

Attestation: This author has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.

Name: Hegang Chen, PhD.

Contribution: This author helped design the study and analyze the data.

Attestation: This author has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.

Name: Christine Wade, MS.

Contribution: This author helped coordinate the study data collection.

Attestation: This author has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.

Name: Lauren Hartsky, MA.

Contribution: This author helped conduct the study and coordinate research.

Attestation: This author has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.

Name: Catriona Miller, PhD.

Contribution: This author helped conduct the study and coordinate research.

Attestation: This author has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.

Name: Cristina Imle, MS.

Contribution: This author helped conduct the study and collect data.

Attestation: This author has seen the original study data and approved the final manuscript.

Name: Raymond Fang, MD.

Contribution: This author helped design the study and provided project oversight.

Attestation: This author has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.

Name: Colin F. Mackenzie, MBChB.

Contribution: This author helped design the study, analyze the data, write the manuscript, and provide project oversight.

Attestation: This author has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.

This manuscript was handled by: Maxime Cannesson, MD, PhD.

Back to Top | Article Outline

ACKNOWLEDGMENTS

The authors thank the CORE Research Staff; the clinicians of the R Adams Cowley Shock Trauma Center; Betsy Kramer, MS; Melissa Binder, MS; and the ONPOINT investigators group, without whose help the data collection and Trauma Registry review for this study would not have been possible. The views expressed in this article are those of the authors and do not necessarily reflect the official policy or position of the Air Force, the Department of Defense, or the U.S. Government. We thank Masimo Corporation for installing the equipment necessary to collect the SpHb data and for providing the SpHb trends. No financial support of investigators or supplies was provided by Masimo Corporation. The ONPOINT investigators group includes Amechi Anazodo, John Blenko, Chein-I Chang, Theresa Dinardo, Joseph duBose, Raymond Fang, Yvette Fouche, Linda Goetz, Tom Grissom, Victor Guistina, George Hagegeorge, Anthony Herrera, John Hess, Peter Hu, Cristine Imle, Matthew Lissauer, Colin Mackenzie, Jay Menaker, Karen Murdock, Sarah Saccicchio, Thomas Scalea, Stacy Shackelford, Robert Sikorski, Lynn Smith, Lynn Stansbury, Deborah Stein, and Chris Stephens.

Back to Top | Article Outline

FOOTNOTES

a Masimo Corporation. Accuracy of Noninvasive and Continuous Hemoglobin Measurement by Pulse CO-Oximetry: Data Submitted by Masimo as Part of FDA 510(k) Clearance. Retrieved from: www.masimo.com/pdf/whitepaper/LAB7131A.pdf. Accessed August 23, 2014.
Cited Here...

b XN-Series analyzers have HGB (g/dL) precision at 3 levels: low, normal, and high, with means 6.5, 13.49, and 17.3, and total coefficient of variation 1.08, 0.62, and 0.89. 510(k) Substantial Equivalence Determination Decision Summary. Retrieved from: www.accessdata.fda.gov/cdrh_docs/reviews/K112605.pdf. Accessed May 25, 2015.
Cited Here...

Back to Top | Article Outline

REFERENCES

1. Holcomb JB. Optimal use of blood products in severely injured trauma patients. Hematology Am Soc Hematol Educ Program. 2010;2010:465–9
2. Perkins JG, Beekley ACSavitsky E, Eastbridge B. Damage control resuscitation. In: Combat Casualty Care: Lessons Learned from OEF and OIF. 2012 Washington, DC Office of the Surgeon General, Borden Institute:121–64
3. Wilson MT, Reeder BJ. Oxygen-binding haem proteins. Exp Physiol. 2008;93:128–32
4. Reisner A, Shaltis PA, McCombie D, Asada HH. Utility of the photoplethysmogram in circulatory monitoring. Anesthesiology. 2008;108:950–8
5. Allen J. Photoplethysmography and its application in clinical physiological measurement. Physiol Meas. 2007;28:R1–39
6. Miller RD, Ward TA, Shiboski SC, Cohen NH. A comparison of three methods of hemoglobin monitoring in patients undergoing spine surgery. Anesth Analg. 2011;112:858–63
7. Colquhoun DA, Forkin KT, Durieux ME, Thiele RH. Ability of the Masimo pulse CO-Oximeter to detect changes in hemoglobin. J Clin Monit Comput. 2012;26:69–73
8. Berkow L, Rotolo S, Mirski E. Continuous noninvasive hemoglobin monitoring during complex spine surgery. Anesth Analg. 2011;113:1396–402
9. Bland JM, Altman DG. Measuring agreement in method comparison studies. Stat Methods Med Res. 1999;8:135–60
10. Clarke WL, Cox D, Gonder-Frederick LA, Carter W, Pohl SL. Evaluating clinical accuracy of systems for self-monitoring of blood glucose. Diabetes Care. 1987;10:622–8
11. Morey TE, Gravenstein N, Rice MJ. Let’s think clinically instead of mathematically about device accuracy. Anesth Analg. 2011;113:89–91
12. Vandromme MJ, Griffin RL, Kerby JD, McGwin G Jr, Rue LW III, Weinberg JA. Identifying risk for massive transfusion in the relatively normotensive patient: utility of the prehospital shock index. J Trauma. 2011;70:384–8
13. Olaussen A, Blackburn T, Mitra B, Fitzgerald M. Review article: shock index for prediction of critical bleeding post-trauma: a systematic review. Emerg Med Australas. 2014;26:223–8
14. Kahraman S, Dutton RP, Hu P, Xiao Y, Aarabi B, Stein DM, Scalea TM. Automated measurement of “pressure times time dose” of intracranial hypertension best predicts outcome after severe traumatic brain injury. J Trauma. 2010;69:110–8
15. Fawcett T ROC Graphs: Notes and Practical Considerations for Data Mining Researchers. Report No. HPL-2003-4. 2003 Palo Alto, CA HP Laboratories
16. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics. 1988;44:837–45
17. Hosmer DW Jr, Lemeshow S, Sturdivant RX Applied Logistic Regression. 20133rd ed. Hoboken, NJ Wiley
18. Coquin J, Dewitte A, Manach YL, Caujolle M, Joannes-Boyau O, Fleureau C, Janvier G, Ouattara A. Precision of noninvasive hemoglobin-level measurement by pulse co-oximetry in patients admitted to intensive care units for severe gastrointestinal bleeds. Crit Care Med. 2012;40:2576–82
19. Nguyen BV, Vincent JL, Nowak E, Coat M, Paleiron N, Gouny P, Ould-Ahmed M, Guillouet M, Arvieux CC, Gueret G. The accuracy of noninvasive hemoglobin measurement by multiwavelength pulse oximetry after cardiac surgery. Anesth Analg. 2011;113:1052–7
20. Frasca D, Dahyot-Fizelier C, Catherine K, Levrat Q, Debaene B, Mimoz O. Accuracy of a continuous noninvasive hemoglobin monitor in intensive care unit patients. Crit Care Med. 2011;39:2277–82
21. Joseph B, Hadjizacharia P, Aziz H, Snyder K, Wynne J, Kulvatunyou N, Tang A, O’Keeffe T, Latifi R, Friese R, Rhee P. Continuous noninvasive hemoglobin monitor from pulse ox: ready for prime time? World J Surg. 2013;37:525–9
22. Lamhaut L, Apriotesei R, Combes X, Lejay M, Carli P, Vivien B. Comparison of the accuracy of noninvasive hemoglobin monitoring by spectrophotometry (SpHb) and HemoCue® with automated laboratory hemoglobin measurement. Anesthesiology. 2011;115:548–54
23. Macknet MR, Allard M, Applegate RL II, Rook J. The accuracy of noninvasive and continuous total hemoglobin measurement by pulse CO-Oximetry in human subjects undergoing hemodilution. Anesth Analg. 2010;111:1424–6
24. Kim SH, Lilot M, Murphy LS, Sidhu KS, Yu Z, Rinehart J, Cannesson M. Accuracy of continuous noninvasive hemoglobin monitoring: a systematic review and meta-analysis. Anesth Analg. 2014;119:332–46
25. Gayat E, Aulagnier J, Matthieu E, Boisson M, Fischler M. Non-invasive measurement of hemoglobin: assessment of two different point-of-care technologies. PLoS One. 2012;7:e30065
26. Naftalovich R, Naftalovich D. Error in noninvasive spectrophotometric measurement of blood hemoglobin concentration under conditions of blood loss. Med Hypotheses. 2011;77:665–7
27. Allard CB, Scarpelini S, Rhind SG, Baker AJ, Shek PN, Tien H, Fernando M, Tremblay L, Morrison LJ, Pinto R, Rizoli SB. Abnormal coagulation tests are associated with progression of traumatic intracranial hemorrhage. J Trauma. 2009;67:959–67
28. Joseph B, Aziz H, Zangbar B, Kulvatunyou N, Pandit V, O’Keeffe T, Tang A, Wynne J, Friese RS, Rhee P. Acquired coagulopathy of traumatic brain injury defined by routine laboratory tests: which laboratory values matter? J Trauma Acute Care Surg. 2014;76:121–5
29. Krausz MMRabinovici R, Frankel HL, Kirton O. Fluid resuscitation for the trauma patient. In: Trauma, Critical Care and Surgical Emergencies. 2010 London, UK Informa Healthcare:1–7
30. Galvagno SM Jr, Hu P, Yang S, Gao C, Hanna D, Shackelford S, Mackenzie C. Accuracy of continuous noninvasive hemoglobin monitoring for the prediction of blood transfusion in trauma patients. J Clin Monit Comput. 2015 [epub ahead of print]
31. Hawkinberry DW, Minokadeh AWilson WC, Grande CM, Hoyt DB. Initial circulation assessment and shock prevention. In: Trauma: Emergency Resuscitation, Perioperative Anesthesia, Surgical Management. 2007;Vol 1 New York, NY Informa Healthcare USA:235–46

Supplemental Digital Content

Back to Top | Article Outline
© 2016 International Anesthesia Research Society