Combat casualties are at high risk for hemorrhage due to the severity and complexity of their injuries. In severely injured casualties evacuated from Iraq and Afghanistan, 69% suffered polytrauma (1), with an average of 4.6 injuries per patient (2). Many of these casualties require hemorrhage control, damage control resuscitation (including massive transfusions), and surgery and rapid evacuation to a higher level of care. The ability to continuously monitor patients during this dynamic period is essential. Intermittent measurements of physiologic indicators such as vital signs, indications of perfusion (e.g., lactate, base deficit, urine output) and hemoglobin may not detect acute changes preceding deterioration or provide timely assessment of treatment response.
Continuous noninvasive hemoglobin monitoring may provide insight into the patient's status during this dynamic period. Most research on the accuracy and precision of continuous hemoglobin (SpHb) has been conducted under steady-state conditions. These conditions have included the induction of anesthesia and patients who are not actively bleeding in the intensive care unit or in emergency department (3–5). Additional studies have been conducted during controlled dynamic conditions, such as a fluid bolus, hemodilution, or controlled hemorrhage with concurrent volume resuscitation (6–8). A limited number of studies monitored patients during acute surgical hemorrhage (9–13), with trauma (14–16), or with postoperative blood loss (17, 18). However, among these studies the reported blood loss ranged only from 200 mL to 847 mL (9–12, 19). Only one study has been completed in patients with massive blood loss, and this study occurred during planned surgical procedures (20). No studies using SpHb have been conducted in combat trauma during resuscitation and stabilization.
The purpose of this study is to describe the accuracy (bias), precision (SD), and trending of continuous hemoglobin (SpHb) compared with laboratory hemoglobin (Hb) during the resuscitation and stabilization of combat trauma patients admitted to two US military trauma hospitals in Afghanistan.
PATIENTS AND METHODS
This was a prospective observational study supported by the Joint Combat Casualty Care Research Team and approved with a waiver of consent by the US Army Medical Research and Materiel Command Institutional Review Board. A consecutive convenience sample of combat trauma patients admitted to US military hospitals at Bagram Air Field or Kandahar, Afghanistan from July to September 2010 were included. Subjects were 18 years of age or older, had an admission base deficit of −5 or less, and/or systolic blood pressure of ≤100 mm Hg, and were intubated. The patients, who came directly from the point of injury or were transferred from another facility where damage control resuscitation/surgery had been performed, were enrolled upon arrival in the trauma bay. Patients who had detainee status or who had injuries to their hands that would preclude placement of the pulse oximeter probe were excluded. Routine care of the patients was not altered for this study, and clinicians were blinded to the SpHb results.
Patients were monitored during resuscitation and stabilization in the emergency department, operating room (OR), and intensive care unit (ICU) with a noninvasive pulse oximeter (Masimo Rainbow SET, Masimo, Irvine, CA, Probe Rev E) attached to a Radical-7 Pulse CO-Oximeter (version 22.214.171.124) for SpHb, SpO2, and pulse rate. The perfusion index (PI) was also monitored. The PI, which is an indicator of pulsatile strength, is derived from the pulse oximeter signal and is affected by changes in vascular tone (21). The manufacturer recommends caution interpreting SpHb results when the PI is <1.4. The sensor was positioned on the ring finger of the non-dominant hand (if available). If there was an injury to the extremity/hand the alternate hand was used. The probe was covered with an opaque shield to eliminate any potential error due to extraneous light sources. The placement of the SpHb probe did not circumvent the placement of a pulse oximeter probe for clinical monitoring. Venous or arterial Hb were drawn as routine part of care during resuscitation and analyzed via laboratory (Beckham Coulter Ac-T 5diff and Ac-T or Beckham Coulter Ac-T diff2) or point of care (iStat). All equipment underwent quality control procedures according to laboratory standards. The laboratory or iStat Hb was used as the reference. Additional variables collected included age, sex, mechanism of injury, Injury Severity Score (ISS), interventions (fluids, blood products, vasoactive medications), vital signs every 15 min and before/after the administration of fluids or blood products, and interventions (fluids, blood products, vasoactive medications).
The SpHb data were recorded continuously and stored every 6 s in the Masimo Radical-7 Pulse Co-Oximeter. For analysis, the data were downloaded and averaged over 1 min. The average SpHb in the minute preceding the lab Hb draw was used for analysis. The Bland Altman method accounting for multiple observations (22) was used to evaluate the agreement and precision between the SpHb and Hb. An a priori clinical limit of ± 1 g/dL was applied to the Bland Altman plot. This limit was based on research that suggests that point-of-care devices are generally within 1 g/dL Hb compared with laboratory Hb testing and that Hb paired differences larger than this may lead to an inappropriate transfusion trigger (23, 24). Error grid analysis (25, 26) was used to graphically demonstrate the clinical significance of the differences between methods using a hemoglobin cutoff of 9 g/dL. A four-quadrant plot was used to demonstrate trending over time (27). Data were reported as mean (SD). A P value <0.05 was considered statistically significant.
Data were collected from 24 combat trauma patients. Data pairs from one patient were excluded from further analysis as the laboratory Hb was drawn from an intravenous line and thought by the treating anesthesiologist to be diluted. This decision was supported by the laboratory and SpHb results (Hb 6–8 g/dL vs. SpHb 11.2–12.8 g/dL), with the SpHb thought to be more consistent with the patient's clinical status and preoperative Hb (13.3 g/dL). A summary of patient characteristics from the final sample size of 23 is presented in Table 1. Based on the Injury Severity Score (ISS), six patients (25%) had minor/moderate injuries, seven (29%) had a moderate/severe injury (ISS 15–24), and nine (38%) had severe/critical injury. During the study period, 15 patients received blood/blood products. No data were recorded on blood products administered before the study. For the 10 patients with direct admission, the median time from injury to study monitoring was 82 min, in contrast to the 11 patients who were transferred to a study site (median time from injury to monitoring, 11.2 h). Patients were monitored for the first 2 to 6 h after admission (mean 3.4 h). There were no deaths during the study period.
Forty-nine paired sets of hemoglobin data from 23 patients were analyzed. Each subject contributed between one and five paired sets (median = 2). The Hb ranged from 4.4 g/dL to 15.1 g/dL. The mean laboratory Hb (Coulter or iStat) was 11.0 ± 2.0 g/dL (Coulter, n = 32: 11.3 ± 1.8 g/dL; iStat, n = 17: 10.3 ± 2.2 g/dL). The mean SpHb was 11.5 ± 1.7 g/dL. In four cases, Hb samples were run on both the iStat and the Coulter. The Coulter Hb was significantly higher than the iStat Hb (1.38 ± 0.3 g/dL; 95% CI −1.87, −0.88, P = 0.003). The lab Hb was ≥9 g/dL in 45 (92%) of the pairs; thus an analysis of bias for Hb greater than or less than 9 g/dL was not conducted. The bias (SpHb-Hb) was 0.5 ± 1.8 g/dL (95% CI of bias = 0.03 to 1.0, P = 0.04), with upper and lower limits of agreement of 3.97 g/dL to −2.97 g/dL (Fig. 1), indicating that on average the SpHb tended to overestimate the lab Hb. The underestimation of the laboratory Hb by the iStat values may have introduced increased bias and imprecision of the SpHb-iStat comparison (SpHb-Coulter Hb bias: 0.3 ± 1.6, 95% LOA −2.8 to 3.4 vs. SpHb-iStat Hb bias: 0.8 ± 2.0; 95% LOA −3.1 to 4.7). The error introduced by the iStat was also found in another study of severely injured ICU patients (28), which supports the need to separate out all pairs using the iStat when evaluating the accuracy and precision of the SpHb. A secondary analysis was conducted to determine if the increased bias observed with the iStat was related to its increased use in unstable patients. Using the shock index (SI) as an indicator of hemodynamic instability (29–32), the SI was higher in the cases where the iStat was used (−1.1 ± 0.4) compared with the Coulter (0.9 ± 0.4), but the difference was not significant (P = 0.06). Further, there was no significant relationship (r = 0.04) between the Hb-SpHb bias and the SI, nor was there a significant difference in the use of the iStat versus the Coulter using SI thresholds of 0.9 and 1.0 as indicators of hemodynamic instability (29–32). There was no proportional bias between the mean Hb and the SpHb-Hb bias (R2 = 0.03, P >0.05). Thirty-seven percent of the paired samples had a bias ≤1 g/dL (37% of SpHb-Coulter pairs and 35% of SpHb-iStat pairs, NS) and 53% of the pairs had a bias <1.5 g/dL.
An error grid (Fig. 2) (25, 26) was constructed to demonstrate the clinical importance of the differences between the laboratory (Coulter or iStat) and SpHb values. The error grid has three zones based on the difference between the SpHb and the control Hb value, and allows for visualization of the potential clinical implications of the differences in values. As described by Morey and Rice (25, 26), Zone A contains an isthmus that represents 10% error for Hb values between 6 g/dL and 10 g/dL (an area where a transfusion decision may occur), whereas the areas below 6 g/dL reflect an area where a transfusion will most likely occur and above 10 g/dL where a transfusion is unlikely. Zone C represents potential for a major therapeutic error in the administration of blood, and Zone B lies between Zones A and C, and reflects a potential for therapeutic error that is not as severe as Zone C. Ninety percent (44/49) of the data-pairs were in Zone A (SpHb bias ± 10% Hb) and 10% (5/49) were in Zone B. There were no cases in Zone C. The iStat was the source of the lab value in 4 of 5 of the Zone B readings.
Trend analysis was conducted to describe the concordance in the direction of change in Hb-SpHb relative to the last paired set using a four-quadrant plot (Fig. 3) (27). For this analysis, a central exclusion zone of 1 g/dL was used, as small changes in Hb tend to cause a large random error. As demonstrated in Figure 3, 86% (12/14) of the pairs outside the 1 g/dL exclusion zone had similar directional change, with a coefficient of determination (R2 = 0.51) for the trend. Using an absolute change in Hb of >1 g/dL, a concurrent absolute change in SpHb of >1 g/dL provided the following diagnostic values: sensitivity: 61% (95% CI 32%–86%), specificity 85% (95% CI 55%–98%), positive likelihood ratio 4.0 (95% CI: 1.04–15.36), negative likelihood ratio: 0.45 (95% CI 0.22–0.94), positive predictive value: 80% (95% CI 44%–97%), and a negative predictive value: 69% (41%–89%).
The SpHb signal was present in 4643 of the 6137 min monitored (76%). By subject, the SpHb was present 76 ± 28% (95% CI: 65%–89%); median 93% of the minutes monitored. In three subjects we were unable to acquire a consistent SpHb signal. Field notes related to these three patients all included comments on cold extremities, despite normothermia.
The mean perfusion index (PI) was 2.6 ± 2.8, with 75% of the PI values ≤3.6. The PI was significantly higher when the SpHb signal was present versus lost (2.5 ± 2.6 vs. 0.95 ± 1, P <0.001), with an absolute PI dropout threshold of 0.53. There was a significant relationship between the PI and the bias (rs = 0.5, P <0.001). As the manufacturer recommends caution when the PI is <1.4, a subset analysis of the bias was performed using this threshold. As demonstrated in Figure 1, in the 45 sets with concurrent bias and PI data, when PI was <1.4, the SpHb underestimated the Hb (n = 21 pairs; bias= −0.24 ± 1.8; 95% LOA: −3.7, 3.3). In contrast, when the PI was ≥1.4, the SpHb overestimated the laboratory Hb (n = 24 pairs; bias =1.1 ± 1.4, 95% LOA = −1.6, 3.3). Further exploration of the effect of vasoconstriction (forearm-finger temperature) on the PI and bias is warranted (33), considering the potential use of SpHb monitoring in seriously injured patients under thermally stressful conditions in the field and during aeromedical transport.
There are several limitations in this pilot study. The sample size was small; however, a majority of the patients included were moderately or severely injured as indicated by the ISS. Despite the severity of the injuries of these combat casualties, only four of the data pairs had a Hb <9 g/dL. The study did not collect data on pre-admissions care, but these results likely reflect the use of tourniquets, and the military hypovolemic and blood transfusion practices used during damage control resuscitation (34).
This is the only study to describe continuous SpHb monitoring in combat casualties undergoing resuscitation and surgical care. The population studied is unique as they were cared for under battlefield conditions, with some of the casualties coming directly from the point of injury. This investigation also provides insight into the patient's physiologic status during emergent resuscitation and stabilization.
The results for pairs using only the Coulter Hb (0.3 g/dL ± 1.6 g/dL; 95% LOA: −2.97 to 3.97 g/dL) can be compared to other studies to avoid any potential measurement error associated with the iStat. This bias and precision are similar to studies where the patient is in a dynamic state such as acute surgical hemorrhage (Table 2) (11–14, 20, 27). The studies by Berkow (20), Baulig (14, 18), Moore (35) are most similar to the present study in regards to the acuteness and amount of blood loss. In the present study, the EBL was not recorded, but 15 patients received blood transfusions during the study period (PRBCs: median 2; range: 0–20 units; whole blood 8 units for one patient), which is consistent with other studies reported in Table 2. In 37% of the data pairs the bias was <1 g/dL, and in 53% of the pairs the bias was <1.5 g/dL. The primary difference between the prior studies and the present study may be the urgency of the surgical procedures, with some of the patients in our study in a more dynamic state associated with emergent resuscitative procedures in contrast to planned surgery. Moore's study followed 418 severely injured trauma patients (69% blunt trauma) for 24-h. Compared with this study, 36.6% of the patients in Moore's study received blood transfusions, and there was no difference in bias for patients who did or did not receive a transfusion. Among the 153 patients who received a transfusion, the 95% confidence limits for the lab Hb and the SpHb were 2.97 to −4.62.
The perfusion state of the patients (75% of data pairs obtained with a PI <3.6) may have contributed to the increased bias in this study. The results of this study are consistent with Miller (13) and Kondo (36) studies where there was increased bias with a lower PI. The effect of low perfusion on the ability to monitor is of concern. In our study the SpHb measurements were obtained in 93% of the minutes monitored; however, there were three patients for whom we were not able to obtain a signal. In contrast, in Moore (35) study, the signal was not obtainable in 34% of the readouts. The difference between the studies may reflect the length of time being monitored.
The results of the error grid analysis, in which 90% of the pairs were in Zone A and 10% were in Zone B (Fig. 2) exceed the limits specified by Morey et al. (25), who suggested that 95% of the pairs be in Zone A and 5% in Zone B. Several factors may be associated with these findings including the use of a point-of-care monitor (four of the five pairs in Zone B used the iStat at the value for comparison) and that an acute change in the patient's status may cause a lag between the microcirculatory (as measured by the SpHb) and macrocirculatory Hb levels (37). One potential method to improve the bias and precision of the SpHb measurements is a single calibration based on the initial Hb-SpHb difference (38, 39). It is not known if this recalibration needs to be accomplished under dynamic conditions.
A potential benefit of continuous SpHb is the ability to detect acute changes in Hb and occult bleeding (37). In this study the SpHb had a sensitivity of 61% and specificity of 85% to detect a 1 g/dL change in Hb. In Baulig (14) study of 26 severely traumatized patients undergoing surgery SpHb had a sensitivity of 59% and specificity of 83% to detect a Hb change greater than 0.5 g/dL. The ability of SpHb to accurately detect changes in Hb was demonstrated in the trend analysis for concurrence (Fig. 3). In Colquhoun (27) study of patients undergoing major spine surgery, 94% of the data were in the quadrants indicating correct directional change, and in Tsuei (28) study of ICU patients at risk for bleeding the SpHb-CBC concordance was 60%, in contrast to the 86% concordance in our study that included patients in both the OR and ICU. The trend analysis for concurrence in our study (r2 = 0.51) was comparable to studies under steady-state conditions in patients in the ICU (r2 = 0.41) and OR (r2 = 0.45) (3, 10), during dynamic conditions, including hemodilution (r2 = 0.78), surgical blood loss (r2 = 0.37), boluses (r2 = 0.56) (7, 10, 12), and trauma surgery (r2 = 0.43) (14).
Noninvasive SpHb did not have adequate precision to serve as a sole indicator of the need for transfusion in the unstable trauma patient. Continuous SpHb may be potentially useful in following trends and in detecting acute changes in Hb in combat casualties. The continuous measurement also provides insight into the variability of Hb in patients in a dynamic state (actively bleeding or being resuscitated), and as suggested by the concordance analysis, may provide an early indicator of clinical changes or deterioration. Additional studies are needed to further describe the changes in SpHb under dynamic conditions and to monitor a patient's state during en route transport.
1. Bridges E, Evers K. Wartime critical care air transport. Mil Med
2009; 174 4:370–375.
2. Champion HR, Holcomb JB, Lawnick MM, Kelliher T, Spott MA, Galarneau MR, Jenkins DH, West SA, Dye J, Wade CE, et al. Improved characterization of combat injury. J Trauma
2010; 68 5:1139–1150.
3. 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 10:2277–2282.
4. Gayat E, Bodin A, Sportiello C, Boisson M, Dreyfus JF, Mathieu E, Fischler M. Performance evaluation of a noninvasive hemoglobin monitoring device. Ann Emerg Med
2011; 57 4:330–333.
5. Knutson T, Della-Giustina D, Tomich E, Wills B, Luerssen E, Reynolds P. Evaluation of a new noninvasive device in determining hemoglobin levels in emergency department patients. West J Emerg Med
2013; 14 3:283–286.
6. Macknet M, Norton S, Kimball-Jones P, Applegate R, Martin R, Allard M. Continuous noninvasive measurement of hemoglobin via pulse co-oximetry. Anesth Analg
2007; 105 6:S108–S109.
7. Macknet MR, Allard M, Applegate RL, Rook J. The accuracy of noninvasive and continuous total hemoglobin measurement by pulse CO-oximetry in human subjects undergoing hemodilution. Anesth Analg
2010; 111 6:1424–1426.
8. Kinsky M, Salter M, Daneshvari S, Indrikovs A, George K. Continuous noninvasive hemoglobin: Impact of hemorrhage on volume expansion of crystalloids in humans. Annual Meeting of the American Society of Anesthesiologists
9. Causey MW, Miller S, Foster A, Beekley A, Zenger D, Martin M. Validation of noninvasive hemoglobin measurements using the Masimo Radical-7 SpHb Station. Am J Surg
2011; 201 5:592–598.
10. Vos JJ, Kalmar AF, Struys MM, Porte RJ, Wietasch JK, Scheeren TW, Hendriks HG. Accuracy of non-invasive measurement of haemoglobin concentration by pulse co-oximetry during steady-state and dynamic conditions in liver surgery. Br J Anaesth
2012; 109 4:522–528.
11. 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 (R) with automated laboratory hemoglobin measurement. Anesthesiology
2011; 115 3:548–554.
12. Applegate R, Collier C, Mangus D, Barr S, Macknet M, Hassanian M, Allard M. Evaluation of absolute and trend accuracy of revision G noninvasive and continuous hemoglobin monitoring during major surgery. Anesth Analg
2013; 116 (Suppl 1):S-384.
13. 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 4:858–863.
14. Baulig W, Prassler S, Sulser S, Dambach M, Biro P, Theusinger O. Non-invasive detection of hemoglobin concentration by pulse co-oximetry in severely traumatized patients. Eur J Anaesthesiol
2013; 30 (Suppl 51):50.
15. Engels P, Romanovsky A, Bagshaw S. Noninvasive hemoglobin monitoring in trauma
patients. Am J Respir Crit Care Med
16. Joseph B, Hadjizacharia P, Aziz H, Snyder K, Wynne J, Kulvatunyou N, Tang A, O’Keeffe T, Latifi R, Friese R, et al. Continuous noninvasive hemoglobin monitor from pulse ox: ready for prime time? World J Surg
2013; 37 3:525–529.
17. Nguyen BV, Vincent JL, Nowak E, Coat M, Paleiron N, Gouny P, Ould-Ahmed M, Guillouet M, Arvieux CC, Gueret G, et al. The accuracy of noninvasive hemoglobin measurement by multiwavelength pulse oximetry after cardiac surgery. Anesth Analg
2011; 113 5:1052–1057.
18. Baulig W, Seifert B, Spahn DR, Theusinger OM. Accuracy of non-invasive continuous total hemoglobin measurement by pulse CO-oximetry in severe traumatized and surgical bleeding patients. J Clin Monit Comput
2015; [epub ahead of print].
19. Giraud B, Frasca D, Debaene B, Mimoz O. Comparison of haemoglobin measurement methods in the operating theatre. Br J Anaesth
2013; 111 6:946–954.
20. Berkow L, Rotolo S, Mirski E. Continuous noninvasive hemoglobin monitoring during complex spine surgery. Anesth Analg
2011; 113 6:1396–1402.
21. Whitepaper: Clinical Applications of Perfusion Index. 2007; Available at: http://www.masimo.co.uk/pdf/whitepaper/LAB3410F.pdf
. Accessed Jan 15, 2016.
22. Bland JM, Altman DG. Agreement between methods of measurement with multiple observations per individual. J Biopharm Stat
2007; 17 4:571–582.
23. Bosshart M, Stover JF, Stocker R, Asmis LM, Feige J, Neff TA, Schuepbach RA, Cottini SR, Bechir M. Two different hematocrit detection methods: different methods, different results? BMC Res Notes
24. Gehring H, Hornberger C, Dibbelt L, Rothsigkeit A, Gerlach K, Schumacher J, Schmucker P. Accuracy of point-of-care-testing (POCT) for determining hemoglobin concentrations. Acta Anaesthesiol Scand
2002; 46 8:980–986.
25. Morey TE, Gravenstein N, Rice MJ. Let's think clinically instead of mathematically about device accuracy. Anesth Analg
2011; 113 1:89–91.
26. Rice MJ, Gravenstein N, Morey TE. Noninvasive hemoglobin monitoring: how accurate is enough? Anesth Analg
2013; 117 4:902–907.
27. 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 2:69–73.
28. Tsuei BJ, Hanseman DJ, Blakeman MJ, et al. Accuracy of noninvasive hemoglobin monitoring in patients at risk for hemorrhage. J Trauma Acute Care Surg
2014; 77 (3 Suppl 2):S134–S139.
29. Rady MY, Nightingale P, Little RA, Edwards JD. Shock index: a re-evaluation in acute circulatory failure. Resuscitation
1992; 23 3:227–234.
30. Cannon CM, Braxton CC, Kling-Smith M, Mahnken JD, Carlton E, Moncure M. Utility of the shock index in predicting mortality in traumatically injured patients. J Trauma
2009; 67 6:1426–1430.
31. Apodaca AN, Morrison JJ, Spott MA, Lira JJ, Bailey J, Eastridge BJ, Mabry RL. Improvements in the hemodynamic stability of combat casualties during en route care. Shock
2013; 40 1:5–10.
32. Mutschler M, Nienaber U, Munzberg M, Bouillon B, Maegele M. TraumaRegister DGU: The Shock Index revisited—a fast guide to transfusion requirement? A retrospective analysis on 21,853 patients derived from the Trauma
Register DGU. Crit Care
2013; 17 4:R172.
33. Yamaura K, Nanishi N, Higashi M, Hoka S. Effects of thermoregulatory vasoconstriction on pulse hemoglobin measurements using a co-oximeter in patients undergoing surgery. J Clin Anesth
2014; 26 8:643–647.
34. Butler FK Jr, Blackbourne LH. Battlefield trauma
care then and now: a decade of Tactical Combat Casualty Care. J Trauma Acute Care Surg
2012; 73 (6 Suppl 5):S395–S402.
35. Moore LJ, Wade CE, Vincent L, Podbielski J, Camp E, Junco DD, Radhakrishnan H, McCarthy J, Gill B, Holcomb JB. Evaluation of noninvasive hemoglobin measurements in trauma
patients. Am J Surg
2013; 206 6:1041–1047.
36. Kondo I, Takamiya T, Suzuki N. Influence of peripheral perfusion index on accuracy of noninvasive hemoglobin monitoring (SpHb). Eur J Anaesthesiol
2013; 30 (Suppl 51):48–49.
37. Naftalovich R, Naftalovich D. Error in noninvasive spectrophotometric measurement of blood hemoglobin concentration under conditions of blood loss. Med Hypotheses
2011; 77 4:665–667.
38. Torp K, Aniskevich S, Pai S, Shine T, Peiris P, Crawford C. In-vivo calibration improves accuracy of non-invasive hemoglobin measurements. Paper presented at: Proceedings of the American Society of Anesthesiologists 2012; Washington, DC.
39. Miyashita R, Hirata N, Sugino S, Mimura M, Yamakage M. Improved non-invasive total haemoglobin measurements after in-vivo adjustment. Anaesthesia
2014; 69 7:752–756.