Keywords
noninvasive hemoglobin monitor - invasive hemoglobin monitor - blood loss - arterial
blood gas - laboratory test
Introduction
Neurosurgical procedures may involve massive and rapid blood loss. Decision to measure
the hemoglobin (Hb) level and when to transfuse blood is important and crucial. In
operating room, Hb estimation is largely relied on values obtained from arterial blood
gas (ABG) analysis. Until recently, only invasive monitoring techniques were available
for the estimation of Hb. Noninvasive (NI) Hb monitoring is found to be more efficient,
less expensive, and preferred by patients compared with invasive Hb monitoring.[1] A new NI device, Masimo Spot Hemoglobin Check Device, received the Food and Drug
Administration (FDA) clearance that allows for quick and noninvasive (NI) checking
of total Hb, oxygen saturation (SpO2), pulse rate, and perfusion index (PI). Continuous, NI Hb monitoring provides clinicians
with the trending changes in Hb, and has the potential to alter red blood cell transfusion
decision making. Awada et al's suggestion to add NI monitor to standard monitoring
resulted in decreased utilization of blood products during intraoperative period specifically
in neurosurgical procedures where excessive blood loss is anticipated while facilitating
earlier transfusions.[2] However, it has been observed that the NI Hb monitoring may not have sufficient
accuracy to minimize the need for invasive Hb monitoring, but it may allow continuous
monitoring of Hb and could guide clinicians as to the need for invasive monitoring.[3] In our study, we aimed to find out the accuracy of NI Hb monitor for estimating
Hb values by comparing it with invasive methods such as arterial and laboratory samples.
Methods
This study was conducted after taking approval from institute ethics committee (Ref:
IEC/NP-315/07.08.2015, RP-13/2015). After taking consent, all American Society of
Anesthesiologists (ASA) physical grade I and II adult patients between 18 and 65 years
of either gender undergoing pituitary surgery (transnasal and trans-sphenoidal) were
included. Patients who refused consent, those with a history of peripheral vascular
disease, hemoglobinopathy, and sickle cell disease and patients whose PI on Masimo
mo nitor was < 1.4 were excluded. A day before surgery, pre-anesthetic checkup was
done. Patients were fasted for 8 hours before the scheduled surgery and received pre-medication
with glycopyrrolate 0.2 mg via intramuscular route, 1 hour prior to surgery. Anesthesia
was induced with fentanyl 2 µg/kg and propofol 1.5 to 2 mg/kg. Tracheal intubation
was facilitated with rocuronium 1 mg/kg. Sevoflurane (0.8–1.2 minimum alveolar concentrations)
in a mixture of O2 and N2O (1:2) was used for the maintenance of anesthesia, together with fentanyl 1 µg/kg,
as an intermittent bolus to maintain analgesia. Rocuronium 0.2 mg/kg every 30 minutes
was used intermittently to provide neuromuscular blockade. A blood sample of 1 mL
was obtained using a radial artery catheter immediately after induction of anesthesia,
but before the start of surgery and approximately every hour thereafter till we removed
the arterial line at the end of surgery. Samples were collected for Hb estimation
from the arterial sample (aHb) using ABG analyzer machine (Eschweiler GmbH & Co. KG)
and laboratory (LabHb) using automated hemoglobin analyzer. Simultaneously, the Hb
reading from the NI Hb monitoring was recorded using Masimo Spot Hemoglobin Check
Device, Masimo Spot Hemoglobin Check Device (Pronto; Irvine, California, United States).
Other values displayed on the monitor, such as the PI was also recorded. Hb values
at a PI < 1.4 are not considered reliable as these are not recommended by the manufacturer,
hence not included for analysis. The blood oxygen saturation values from the standard
pulse oximeter (SpO2), NI (SpO2), and ABG were noted. Also, the core body temperature was noted each time the values
were recorded.
Statistical analysis was performed using Stata 12.0 (StataCorp LP, College Station,
Texas, United States). Data were presented as number (percentage) or mean ± standard
deviation (SD) as appropriate. Bland–Altman plot was added to find out the agreement
between Hb values drawn from three different techniques. The p-value < 0.05 was considered statistically significant.
Results
A total of 30 patients participated in the study, which was conducted over a period
of 1 year. None of the patient was excluded from the study. The male to female ratio
was 13:17. The other demographic characteristics including mean age of 40.83 (17.03)
and mean weight of 66.5 (12.31). Hb could be measured up to two time points (Hb1 and
Hb2) in 30 patients, up to three time points (Hb1, Hb2, and Hb3) in 27 patients, and
up to four time points (Hb1, Hb2, Hb3, and Hb4) in only 18 patients. This was due
to the difference in duration of surgery ([Fig. 1]). At different time points, there was a trend, which showed NI Hb monitor with the
highest Hb values followed by Hb values obtained from arterial sample and the laboratory
test ([Fig. 2]).
Fig. 1 Number of patients screened at different time points.
Fig. 2 Hemoglobin estimation at different time points. Hb, hemoglobin; hr, hour.
[Table 1] shows the correlation between different techniques of Hb estimation. ([Fig. 3]) displays the Bland–Altman plot of the relationship between the observed differences
between Hb values of Laboratory and NI Hb monitor and the mean of the two measures.
Limits of agreement (horizontal lines) indicate that 28 of the 30 estimates of NI
Hb values were within the limits. The limits of agreement are defined as the mean
difference ± 2 SD, and the calculated lower and upper limits for Laboratory and NI
were between–4.5 and +2.7. ([Fig. 4]) displays the Bland–Altman plot of the relationship between the observed differences
between Hb values of arterial and NI Hb monitor and the mean of the two measures.
Limits of agreement (horizontal lines) indicate that 28 of the 30 estimates of NI
Hb values were within the limits. The limits of agreement are defined as the mean
difference ± 2 SD, and the calculated lower and upper limits for Laboratory and NI
are between–4.6 and +3.6. ([Fig. 5]) displays the Bland–Altman plot of the relationship between the observed differences
between Hb values of laboratory and arterial and the mean of the two measures. Limits
of agreement (horizontal lines) indicate that 29 of the 30 estimates of arterial Hb
values were within the limits. The limits of agreement are defined as the mean difference
± 2 SD, and the calculated lower and upper limits for laboratory and NI were between–1.6
and +2.4. In ([Fig. 6]), the regression lines show the relation between the laboratory and NI Hb trends.
([Fig. 7]) shows the relationship between the arterial and NI trends, and ([Fig. 8]) shows the relationship between the arterial and laboratory trends, respectively.
Table 1
Correlation between different methods of hemoglobin estimation
|
Total patients
|
Correlation
|
Significance
|
|
Arterial and laboratory
|
30
|
0.7040
|
0.0000
|
|
Laboratory and noninvasive
|
30
|
0.2355
|
0.2103
|
|
Arterial and noninvasive
|
30
|
0.1059
|
0.5775
|
Fig. 3 Bland–Altman plot of correlation between observed differences between laboratory
and NI. Hb, hemoglobin; NI, noninvasive.
Fig. 4 Bland–Altman plot of correlation between observed differences between arterial and
NI. Hb, hemoglobin; NI, noninvasive.
Fig. 5 Bland–Altman plot of correlation between observed differences between arterial and
laboratory. Hb, hemoglobin.
Fig. 6 Scatter diagram showing relationship between laboratory and NI Hb values trends.
Hb, hemoglobin; NI, noninvasive.
Fig. 7 Scatter diagram showing relationship between arterial and NI Hb values trends. Hb,
hemoglobin; NI, noninvasive.
Fig. 8 Scatter diagram showing relationship between arterial and laboratory Hb values trends.
Hb, hemoglobin.
Discussion
Determination of Hb concentration by standard methods is time consuming, invasive,
and intermittent. NI Hb monitoring devices have the potential for detecting sudden
changes in a patient's Hb concentration in blood. This NI method of Hb determination
provides a comfortable environment to the patient as well as reduces the risk of infection,
number of required working personnel, and long-term costs. However, the accuracy of
NI method can be influenced by many clinical factors such as perfusion state, temperature,
a large volume shift, type of infused fluid, and age of the patient. In our study,
we aimed to assess the accuracy of NI method of Hb estimation over invasive methods.
Based on the results of the Bland–Altman plots, in our study, the calculated 95% confidence
interval (CI) for the difference calculated on the laboratory and NI Hb value was–4.5
and +2.7. Therefore, for an actual Hb value of 12 mg/dL, it could be reported to be
either as low as 7.5 mg/dL or as high as 14.7 mg/dL. It is definitely considered to
be clinically significant, and management of patients on the basis of this report
could be hazardous. Again, for arterial and NI Hb values on the basis of same plot,
there would be wide upper and lower Hb range difference which is not acceptable. The
reason for this wide range could be smaller sample size or inaccuracy of NI Hb estimation
method. The correlation coefficient between the NI and laboratory values was 0.235,
and between the NI and ABG analysis values was 0.105, which showed no correlation.
However, Hb values between arterial and laboratory analysis showed a good correlation
(correlation coefficient of 0.707), which might serve as an explanation for inaccuracy
of NI Hb estimation method thereby leading to wide variation in other two paired groups.
As per the result of Scatter diagrams, invasive and NI methods of Hb estimation showed
poor correlation between themselves ([Figs. 6] and [7]). However, the scatter diagram between arterial and laboratory Hb estimation methods
showed good correlation ([Fig. 8]).
Findings from our study suggest that NI Hb monitor can neither replace ABG sampling
(aHb) for Hb estimation nor Hb estimation from laboratory tests. Though both NI and
aHb methods provide immediate Hb values during intraoperative period for Hb estimation,
continuous real time Hb monitoring is the advantage with NI, which is not possible
with aHb since it requires intermittent blood sampling. We also compared both invasive
methods of Hb estimation techniques, LabHb and ABG analysis, and found a good correlation
between the two (correlation coefficient of 0.7040). Hb estimation by NI monitor depends
on the adequacy of blood flow to the finger, which is indirectly reflected by PI.
PI is a calculated value, which is displayed on NI monitor, and Hb values displayed
on monitor with PI values of < 1.4 are not considered reliable. The alteration in
finger perfusion either underestimates or overestimates the true Hb values depending
on the decreased or increased tissue perfusion. So, PI is an important clinical indicator
of actual Hb values on NI monitor. According to Khanna et al, NI Hb monitor does not
have sufficient accuracy to minimize the need for invasive Hb monitoring, which includes
both ABG and laboratory sample.[4] One limitation of their study was that they could not use NI monitor properly in
8 patients out of 30 due to PI value < 1.4. Throughout surgery in our study, all patients
had PI values >1.4. In contrast, Vora and Desai conducted a study that was performed
in the intensive care unit and compared transcutaneously spectroscopically NI measured
Hb values with venous Hb values. They concluded that there is a good relation between
the two methods for measuring Hb. However, authors further concluded that larger studies
are required to validate this NI method in those with conditions that affect the perfusion.[5] Joseph et al observed that the NI Hb monitoring was found to have excellent correlation
with invasive Hb measurement in trauma patients, and its application allows immediate
and accurate Hb measurement.[6] In our study, we found a poor correlation between Hb values by invasive and NI Hb
monitoring methods; however, the main difference between ours and their study was
that they compared LabHb and NI, whereas in our study, we compared two invasive methods
(aHb and LabHb) of Hb estimation with NI Hb estimation method. Another difference
was the patient population that in our study included non-traumatic patients. In another
study by Applegate et al, they compared NI, aHb, and arterial finger stick blood with
LabHb and observed that all three methods provided similar intraoperative guidance
regarding increase or decrease in Hb value, whereas these cannot be used for guide
transfusion decision making.[7] Butwick et al reported that despite a significant correlation between NI and laboratory
Hb values, NI monitor overestimated Hb values compared with laboratory Hb values.[8] We also observed the same trend of higher Hb values with NI compared with laboratory
Hb. However, NI always overestimated Hb values at different time points compared with
arterial Hb values. In our study, LabHb values were the lowest when compared with
NI and aHb values.
Conclusion
From our study, we conclude that NI method of Hb estimation may be successfully used
in clinical practice; however, this method cannot replace the invasive Hb estimation
methods such as ABG analysis and laboratory method of Hb estimation. More trials are
required to find out the accuracy between noninvasive method of Hb estimation and
invasive methods of Hb estimation. Inclusion of NI method of Hb estimation in standard
monitor list could be helpful in instantaneous assessment of blood loss and guiding
blood transfusion therapy in patients at risk of bleeding.
