Secondary Logo

Journal Logo

Monitoring and technology

Accuracy of noninvasive haemoglobin measurement by pulse oximetry depends on the type of infusion fluid

Bergek, Christian; Zdolsek, Joachim H.; Hahn, Robert G.

Author Information
European Journal of Anaesthesiology: December 2012 - Volume 29 - Issue 12 - p 586-592
doi: 10.1097/EJA.0b013e3283592733



Measurement of haemoglobin concentration (Hb) is usually made in blood sampled from a cubital vein. During surgery, Hb is essential for decisions about whether to transfuse erythrocytes. Therefore, sporadic measurement of Hb is performed during major surgery as a guide to when transfusion should be initiated.

The Hb concentration can also be inferred noninvasively and continuously by multiwavelength pulse co-oximetry (SpHb), which might serve as an attractive alternative to invasive sampling. This recently reviewed technique1 has been used in studies involving volunteers and surgery, with varying conclusions about accuracy and precision.2–11 However, little is known about how SpHb reacts to specific procedures performed during surgery, such as intravenous infusion of crystalloid and colloid fluid. There is some evidence that SpHb changes more than Hb during volume loading with Ringer's acetate7 but no evaluation of the accuracy of SpHb during administration of colloid fluid has been performed.

In the present study, we investigated the reliability (accuracy and precision) of SpHb to measure Hb during and after infusion of Ringer's acetate and 6% hydroxyethyl starch 130/0.4 in volunteers. These fluids were given separately and in combination because crystalloid and colloid fluids are often administered together in major surgery. The hypothesis was that both infusions would change the accuracy of SpHb as a measure of Hb. The perfusion index was also recorded because SpHb seemed to provide lower values when perfusion index is low.2


Ten male volunteers aged between 18 and 28 (mean 22) years and with a body weight between 65 and 101 (mean 79) kg underwent three infusion experiments between August and December 2010. The protocol was part of a project evaluating the blood-volume expanding effects of mixed fluid therapy (starch solution followed by crystalloid solution). The study was approved by the Regional Ethics Committee, Karolinska Institutet, 171 77 Stockholm, Sweden, on 18 September 2009 (Dnr 2009/1091–31/2; Chairperson Ulla Erlandsson) and registered at ClincalTrials by identifier NCT01195025. Each volunteer gave his consent for participation after being informed about the study both orally and in writing.

The experiments started between 07.30 and 08.30 h in the Department of Intensive Care at Linköping University Hospital. The volunteers had fasted since midnight but were allowed to eat one sandwich and drink one glass (200 ml) of liquid at 06.00 h. When they arrived at the department, they rested on a bed below an OPN Thermal Ceiling radiant warmer (Aragon Medical, River Vale, New Jersey, USA) placed about 1 m above them. The heat was adjusted to achieve maximum comfort. A cannula was placed in the antecubital vein of each arm to infuse fluid and sample blood, respectively. Thirty minutes of rest to reach haemodynamic steady state was allowed before the experiments started.


Each volunteer underwent the following three experiments in random order, separated by at least 7 days (Fig. 1):

  1. Infusion of Ringer's acetate 20 ml kg−1 over 30 min (two received 25 ml kg−1);
  2. Infusion of starch 10 ml kg−1 over 30 min;
  3. Infusion of combined starch and Ringers’ acetate; 10 ml kg−1 of starch was infused between 0 and 30 min, followed by 20 ml kg−1 of Ringer's acetate between 105 and 135 min.
Fig. 1
Fig. 1:
No captions available.

The crystalloid fluid was acetated Ringer's (Baxter, Deerfield, Illinois, USA; sodium 130 mmol l−1, chloride 110 mmol l−1, acetate 30 mmol l−1, potassium 4 mmol l−1, calcium 2 mmol l−1, and magnesium 1 mmol l−1). The colloid was hydroxyethyl starch 6% 130/0.4 (Voluven, Fresenius Kabi; Bad Homburg, Germany; sodium 154 mmol l−1, chloride 154 mmol l−1).

The volume of colloid was chosen to correspond to previous work using albumin.12 The infused volume of Ringer's acetate has usually been 25 ml kg−1 in studies of plasma volume expansion in healthy volunteers13 but we reduced it slightly to avoid excessive cardiovascular strain in the combined experiment.

The fluids were administered at room temperature (23oC) via infusion pumps (Volumat MC Agilia, Fresenius Kabi).


Venous blood (3–4 ml) was withdrawn from an antecubital venous cannula, using a vacuum tube, without stasis of the upper arm. The baseline sample was drawn in duplicate, and the mean was used in further calculations. A small volume of blood was drawn before each sample, and the volume replaced by 2 ml of 0.9% isotonic saline to prevent clotting. The venous blood Hb was measured on a Cell-Dyn Sapphire haematology cell counter (Abbot Diagnostics, Abbot Park, Illinois, USA). Duplicate samples drawn at baseline ensured a coefficient of variation of 1.2%.

The sampling intensity varied slightly depending on the length of the experiment. In the Ringer experiment, blood was drawn every 5 min up to 60 min, and thereafter every 10 min up to 180 min. The same protocol was followed when starch was infused alone, but the follow-up continued with blood sampling every 30 min up to 420 min. In the combined experiment, the higher sampling intensity (every 5 min) was reinstituted for 60 min when the second infusion started.

SpHb and perfusion index were measured by the Radical 7 pulse CO-oximeter (Masimo Corp., Irvine, California, USA) which uses light of multiple wavelengths and also advanced filtering and processing of the signals to yield these values.1 A single-use adhesive sensor of type R2–25a was attached to the middle finger of one hand. The venous samples were drawn from the same arm and the infusions were given in the other. The software delivered by the manufacturer was SET V7.6.0.1. The data were averaged every 8 s.

Perfusion index is a measure of the pulse amplitude in the finger and is derived from the ratio between the pulsatile and the nonpulsatile absorption of infrared light. For each invasive blood sample, we recorded the SpHb and perfusion index displayed on the front of the Radical 7. Haemodynamic monitoring also included noninvasive blood pressure measurement and heart rate.


Data are presented as the median and 25th to 75th percentiles, except where indicated. Differences between paired data were evaluated by the Wilcoxon matched-pair test, and differences between unpaired samples by the Mann–Whitney U test. Comparisons among three groups were made by applying the Kruskal–Wallis test.

The influence of several factors on a continuous variable was tested by two-way ANOVA, and the relationship between variables by simple and multiple linear regression (where r = correlation coefficient).

The accuracy (bias) of using SpHb to indicate Hb was expressed as the absolute or relative difference between the paired measurements, the latter being:

The precision of using SpHb to indicate Hb was expressed as the absolute value of the relative difference between the paired measurements. The 95% prediction interval for the absolute difference between SpHb and Hb is the range in which 95% of the SpHb−Hb differences are to be found. The accuracy and precision of SpHb to indicate Hb are also illustrated by Bland–Altman plots. Calculations were considered statistically significant if P was less than 0.05.


All 10 volunteers completed the study, which comprised 30 experiments altogether (Fig. 1). Baseline data are shown in Table 1.

Table 1
Table 1:
Data on volunteers, baseline values

Haemoglobin changes during infusion

At the end of the first infusion of Ringer's acetate, SpHb had decreased more than Hb (15 vs. 8%; P < 0.005; n = 10; Fig. 2 a). At the end of the infusions of starch, SpHb had decreased less than Hb (7 vs. 11%; P < 0.02; n = 20; Fig. 2 b and c). At the end of the infusion of Ringer's acetate in the combined experiment, SpHb had again decreased more than Hb (11 vs. 5%; P < 0.03; n = 10; Fig. 2 c).

Fig. 2
Fig. 2:
No captions available.

Differences between SpHb and haemoglobin

SpHb tended to be lower than Hb at baseline (median 136.5 vs. 144 g l−1; P = 0.07; n = 30) which yielded a bias of −3.8% and a precision of 6.6%. Infusion of Ringer's acetate increased the bias by 7.0 (2.5 to 11.1)% (P < 0.04, median, 25th to 75th percentile range) whereas starch decreased the bias by 4.3 (0.9 to 7.2)% (P < 0.02; n = 20; Table 2). The precision had become 4.6 (−1.7 to 6.1)% poorer at the end of the Ringer infusions (P < 0.02) whereas starch did not affect the precision, the median change being 0.8 (−3.5 to 3.7%) (not statistically significant, Table 2).

Table 2
Table 2:
The noninvasive Hb (SpHb) and invasive venous Hb concentrations and the accuracy and precision of SpHb to predict invasive Hb in the course of the three infusion experiments. The perfusion index is also shown

The mean difference between all 956 pairs of SpHb and Hb (the bias) was −0.7 g l−1. The median (25th to 75th percentiles) were −2 (−10 to 8) g l−1, which corresponds to an accuracy of −0.8 (−7.5 to 5.9)% and a median precision of 6.6 (3.1 to 10.7)% (Table 2). The 95% prediction interval for the SpHb–Hb difference ranged from −24.9 to 23.7 g l−1 (Figs 3 and 4). Two-way ANOVA showed that the bias was dependent on the infusion experiment, but also strongly dependent on the volunteer (each factor P < 0.001; Table 3). The difference between SpHb and Hb became more positive (so that SpHb > Hb) at lower Hb concentrations (r = 0.42, P < 0.001; Fig. 4).

Fig. 3
Fig. 3:
No captions available.
Fig. 4
Fig. 4:
No captions available.
Table 3
Table 3:
The accuracy (bias) of SpHb in indicating invasive Hb concentration during infusion experiments, expressed as the mean relative difference for all data in each of 10 volunteers

Perfusion index

Perfusion index was 7.0 (4.3 to 9.2)% at baseline. The Ringer infusions decreased perfusion index from (median) 7.0 (4.4 to 11.0)% to 2.5 (1.3 to 6.4)% and the starch infusion from 5.4 (5.0 to 8.1)% to 3.0 (1.9 to 6.1)% (repeated-measures ANOVA P < 0.01 and 0.02, respectively; Table 2).

The SpHb–Hb difference increased with a higher perfusion index. Thus, a median perfusion index above 7.0% during the experiments was associated with a positive bias whereas for a perfusion index of 7.0% or less the average bias was negative (Table 4). This was explained by the fact that SpHb but not Hb increased with perfusion index (Fig. 5). Multiple regression analysis showed that the effect of perfusion index on the bias was statistically independent of the opposite effect of Hb per se; this is illustrated in Fig. 4 (combined factors; r = 0.49; P < 0.001).

Table 4
Table 4:
Differences in the accuracy (bias) and precision of noninvasive SpHb in indicating invasive Hb during infusion experiments in 10 volunteers, depending on the perfusion index
Fig. 5
Fig. 5:
No captions available.


The results show that the accuracy of noninvasive SpHb is dependent on the type of infusion fluid administered. Starch caused SpHb to decrease much less than Hb, and the difference was long lasting. In contrast, when infusing Ringer's acetate, the decrease in SpHb was greater and more transient. Both of these effects could also be discerned in the combined experiment (Fig. 2).

Other factors were also found to affect the bias of the SpHb measurement. In addition to the choice of infusion fluid, the between-volunteer variation was important (Table 3). A low Hb concentration promoted a more positive bias, which means that SpHb overestimated Hb when a correct indication could be clinically important. Moreover, SpHb increased slightly with perfusion index, which confirms previous findings.2 Other authors have reported a more negative SpHb–Hb difference when perfusion index is below 2%3 whereas our findings were that the same relationship exists also when perfusion index exceeds 7% (Table 4).

Starch administration, low Hb and high perfusion index were all factors which promoted a more positive SpHb–Hb difference, thereby acting to mislead the clinician with regard to the need for erythrocyte transfusion. The reasons for inconsistency of the SpHb measurement are unclear. One hypothesis is that our fluid load might have disturbed this balance by expanding the vessels and also by diluting Hb differently in arterioles, capillaries, and veins.15 Faster disappearance of crystalloids than colloids from the bloodstream16,17 promotes tissue oedema, which could diminish the relative strength of the pulsatile part of the signal by affecting the background noise. The time-course of the negative SpHb–Hb difference when acetated Ringer's is infused (Fig. 2a) is consistent with findings that the net movement of fluid from plasma to the interstitium in the arm is reversed within 2.5 min after the end of a brisk infusion.18 In contrast, accumulation of infused fluid in peripheral tissues is more long lasting when the whole body is studied.13 The opposite change of the SpHb–Hb difference in response to starch might possibly be due to overlapping of physical absorbance characteristics of starch and Hb. Experimental studies investigating the microvasculature in combination with spectrophotometry and optical physics may give answers to the open questions in this study.

The bias calculated from a large number of paired measurements is usually quite small in studies using the SpHb technology. The bias when infusing crystalloid fluid was −1.6% in a previous study,7 and in blood withdrawal followed by crystalloid fluid, the value of the bias averaged −1.5 g l−1.9 In spinal surgery, the bias was −2.6,2 −310 and −12.711 g l−1. Other data from the perioperative setting were −0.28, −2.94 and 05 g l−1. Extreme biases include −13 to −17 g l−1 in cardiac intensive care3 and −18 g l−1 in emergency care.6 In all these studies, the SpHb had a lower mean value than Hb, that is a negative bias.

In contrast, poor precision is a problem. According to published graphs and charts, the 95% prediction interval for the SpHb–Hb difference has been 404,9,10, 40–508, 67–802,3,11, and 1106 g l−1. In the present study, the range was almost 50 g l−1. Some of the wide variability in the SpHb–Hb difference in previous studies may have been caused by the fluid therapy used when data was collected.

The SpHb at baseline differed slightly in the three experiments (Table 2). Small variations can be due to the state of hydration of the volunteers because the Radical 7 measures SpHb in arterial blood whereas the sampled blood was venous. The arteriovenous Hb difference has been reported to be −1.8 g l−1 in nonfasting volunteers in the afternoon,19 but amounted to only −0.3 g l−1 in semi-fasting male volunteers studied in the morning.18 The gradient might even become zero or slightly positive after an overnight fast20 because of fluid transport away from the arm. The Radical 7 can be set to reflect SpHb in venous blood, which simply makes it report values which are 0.7 to 1.0 g l−1 higher. This option was not used in the present study because the arteriovenous Hb difference was likely to be close to zero at baseline. However, the true difference was probably increased to approximately −1 g l−1 during the infusion of Ringer's acetate.18 The arteriovenous Hb difference induced by the infusion can, therefore, only explain a fraction of the difference between SpHb and Hb at the end of the Ringer's infusions, which amounted to almost −11 g l−1.

In a previous report, perfusion index decreased when Ringer's acetate was infused,7 and the same change was seen in the present study. This effect is surprising because perfusion index is an index of blood flow, which is expected to increase as a result of volume loading. As this did not occur, we suspected that the oedema caused by the infusion had decreased the ratio between pulsatile and nonpulsatile flow. However, this cannot be correct because perfusion index also decreased when starch was infused, despite the fact that colloid fluids are not deposited extravascularly early on during an infusion.16 The fluids infused at room temperature cooled the body and this may have resulted in vasoconstriction and less perfusion. However, as all fluids were of the same temperature and the amount of Ringer's was twice that of starch, the cooling effect would then logically be twice as large in the Ringer's experiment, but the fall in perfusion index was much greater when infusing starch. Therefore, temperature change does not provide a satisfactory explanation for the change in perfusion index.

Limitations of our study include the fact that the infusions were administered to normovolaemic individuals, which is often not the case in clinical practice; the difference between SpHb and Hb might be different when a hypovolaemic patient is infused. Moreover, differences in healthy volunteers may not completely reflect the situation in anaesthetised patients. Moreover, only male volunteers were studied because females have difficulty voiding in the supine position.

In conclusion, changes in SpHb in response to intravenous infusion of fluid differ depending on whether a crystalloid (Ringer's acetate) or colloid (starch solution) fluid is given. Starch administration, low Hb concentration and high perfusion index are all factors which promote a positive SpHb–Hb difference, which can mislead the clinician to underestimate the need for erythrocyte transfusions. However, infusion of Ringer's solution, a high blood Hb and a low perfusion index act to exaggerate the need for transfusing erythrocytes.


Assistance with the study: assistance with the study was given by nurse anaesthetists Susanne Lind and Helen Didriksson. The Intensive Care Unit in Linköping University Hospital provided us with room for the experiments.

Financial support and sponsorship: financial support was received from Stockholm City and Östergötland County Councils.

Conflicts of interest: RH has provided lectures about fluid therapy sponsored by Baxter Medical Corp. There are no other conflicts of interest.


1. Shamir MY, Avramovich A, Smaka T. The current status of continuous noninvasive measurement of total, carboxy, and methemoglobin concentration. Anesth Analg 2012; 114:972–978.
2. Miller RD, Ward TA, Shiboski SC, Cohen NH. A comparison of three methods of hemoglobin monitoring in patients undergoing spine surgery. Anesth Analg 2011; 112:858–863.
3. Nguyen B-V, Vincent J-L, Nowak E, et al. The accuracy of noninvasive hemoglobin measurement by multiwavelength pulse oximetry after cardiac surgery. Anesth Analg 2011; 113:1052–1057.
4. Causey MW, Miller S, Foster A, et al. Validation of noninvasive hemoglobin measurements using the Masimo Radical-7 SpHb Station. Am J Surg 2011; 201:592–598.
5. Frasca D, Dahyot-Fizelier C, Catherine K, et al. Accuracy of a continuous noninvasive hemoglobin monitor in intensive care unit patients. Crit Care Med 2011; 39:2277–2282.
6. Gayat E, Bodin A, Sportiello C, et al. Performance evaluation of a noninvasive hemoglobin monitoring device. Ann Emerg Med 2011; 57:330–333.
7. Hahn RG, Li Y, Zdolsek J. Noninvasive monitoring of blood haemoglobin for analysis of fluid volume kinetics. Acta Anaesthesiol Scand 2010; 54:1233–1240.
8. Lamhaut L, Apriotesei R, Combes X, et al. Comparison of the accuracy of noninvasive hemoglobin monitoring by spectrophotometry (SpHb) and HemoCue with automated laboratory hemoglobin measurement. Anesthesiology 2011; 115:548–554.
9. 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; 111:1424–1426.
10. Berkow L, Rotolo S, Mirski E. Continuous noninvasive hemoglobin monitoring during complex spine surgery. Anesth Analg 2011; 113:1396–1402.
11. 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.
12. Hedin A, Hahn RG. Volume expansion and plasma protein clearance during intravenous infusion of 5% albumin and autologous plasma. Clin Sci 2005; 106:217–224.
13. Drobin D, Hahn RG. Kinetics of isotonic and hypertonic plasma volume expanders. Anesthesiology 2002; 96:1371–1380.
14. Nadler SB, Hidalgo JU, Bloch T. Prediction of blood volume in normal human adults. Surgery 1962; 51:224–232.
15. Naftalovich R, Naftalovich D. Error in noninvasive spectrophotometric measurement of blood hemoglobin concentration under conditions of blood loss. Med Hypotheses 2011; 77:665–667.
16. Ueyama H, He YL, Tanigami H, et al. Effects of crystalloid and colloid preload on blood volume in the parturient undergoing spinal anesthesia for elective Cesarean section. Anesthesiology 1999; 91:1571–1576.
17. McIlroy DR, Kharasch ED. Acute intravascular volume expansion with rapidly administered crystalloid or colloid in the setting of moderate hypovolemia. Anesth Analg 2003; 96:1572–1577.
18. Svensen CH, Rodhe PM, Olsson J, et al. Arteriovenous differences in plasma dilution and the distribution kinetics of lactated ringer's solution. Anesth Analg 2009; 108:128–133.
19. Yang Z-W, Yang S-H, Chen L, et al. Comparison of blood counts in venous, fingertip and arterial blood and their measurement variation. Clin Lab Haem 2001; 23:155–159.
20. Hahn RG, Lindahl CC, Drobin D. Volume kinetics of acetated Ringer's solution during experimental spinal anaesthesia. Acta Anaesthesiol Scand 2011; 55:987–994.

colloid; crystalloid; haemoglobin; pulse oximetry

© 2012 European Society of Anaesthesiology