Duplicate measures of hemoglobin mass within an hour: feasibility, reliability, and comparison of three devices in supine position

Abstract Duplicate measure of hemoglobin mass by carbon monoxide (CO)-rebreathing is a logistical challenge as recommendations prompt several hours between measures to minimize CO-accumulation. This study investigated the feasibility and reliability of performing duplicate CO-rebreathing procedures immediately following one another. Additionally, it was evaluated whether the obtained hemoglobin mass from three different CO-rebreathing devices is comparable. Fifty-five healthy participants (22 males, 23 females) performed 222 duplicate CO-rebreathing procedures in total. Additionally, in a randomized cross-over design 10 participants completed three experimental trials, each including three CO-rebreathing procedures, with the first and second separated by 24 h and the second and third separated by 5–10 min. Each trial was separated by >48 h and conducted using either a glass-spirometer, a semi-automated electromechanical device, or a standard three-way plastic valve designed for pulmonary measurements. Hemoglobin mass was 3 ± 22 g lower (p < 0.05) at the second measure when performed immediately after the first with a typical error of 1.1%. Carboxyhemoglobin levels reached 10.9 ± 1.3%. In the randomized trial, hemoglobin mass was similar between the glass-spirometer and three-way valve, but ∼6% (∼50 g) higher for the semi-automated device. Notably, differences in hemoglobin mass were up to ∼13% (∼100 g) when device-specific recommendations for correction of CO loss to myoglobin and exhalation was followed. In conclusion, it is feasible and reliable to perform two immediate CO-rebreathing procedures. Hemoglobin mass is comparable between the glass-spirometer and the three-way plastic valve, but higher for the semi-automated device. The differences are amplified if the device-specific recommendations of CO-loss corrections are followed.

Nonetheless, performing duplicate measurements several hours apart constitutes a logistical challenge.To attempt overcoming such practical challenge while maintaining measurement precision, the primary aim of the present project was to evaluate the feasibility and reliability of performing immediate duplicate CO-rebreathing measurements.
Currently, two commercially available devices designed for CO-rebreathing are available; a glass spirometer (Blood tec GmbH, Bayreuth, Germany) and a semi-automated electromechanical device (Detalo Health, Birkerød, Denmark).Both devices have been applied in science [1][2][3][4][5][6][8][9][10][26][27][28], but a recent report indicate that the semi-automated device estimates a 10% higher total hemoglobin mass compared to the glass-spirometer [29].The authors ascribe some of the discrepancy (~3%) to differences in body posture (i.e.seated vs. supine).However, a subsequent study reported that body posture was unlikely to account for the difference [30].Accordingly, a secondary aim of the present study was to establish whether the total hemoglobin mass obtained by red cell mass; carboxyhemoglobin; erythrocyte volume; reliability and validity; hematology; study; methodological; blood volume glass spirometer is comparable to the semi-automated device when obtained in the same body posture or whether a discrepancy between the devices per se exists.In addition, it was investigated whether a standard plastic three-way valve designed for pulmonary measurements is a suitable device for CO-rebreathing in comparison to the two established devices.A three-way plastic valve has the potential advantage of being less fragile than the glass spirometer, less expensive than the semi-automated device, provide an access to room air while the user is fully mounted to the device prior to initiation of the rebreathing, while also being useful for unrelated pulmonary measurements.
In summary, it was hypothesized that measuring total hemoglobin mass in immediate, consecutive duplicates provides a typical error <2% and a similar result as compared to measuring total hemoglobin mass once on two consecutive days interspersed by 24 h.Furthermore, it was hypothesized that measuring total hemoglobin mass using a semi-automated device provides a higher total hemoglobin mass in comparison to using a glass spirometer or a standard three-way plastic valve when measured in the supine position.

Immediate duplicate measurements
To investigate whether duplicate measures of total hemoglobin mass performed immediately following one another is a feasible and reliable strategy, data from two published [7,19] and four yet unpublished studies were combined.The combined data included 55 healthy subjects (22 men, 23 women, with an age, height, and weight of 27 ± 7 years, 178 ± 8 cm and 73 ± 11 kg, respectively) being subjected to a total of 222 immediate duplicate measures of total hemoglobin mass.
These duplicate measurements were performed 5-10 min apart using a glass-spirometer (n = 10), a semi-automated device (n = 82) or a three-way plastic valve (n = 130) in accordance with the details provided in the following method sections.
The ethics committee of Copenhagen, Denmark approved all the published (H-19021948, H-18013069) and yet unpublished (H-21022277, H-21011041, H-21075850) studies.All subjects were informed orally and in writing about potential risks and discomforts associated with participation before written consent was obtained.

Device comparison
To investigate whether CO-rebreathing procedures with the glass-spirometer, the semi-automated device and the three-way valve provide a similar total hemoglobin mass, ten non-smoking participants (9 males, 1 female) with a mean± SD age, height, and weight of 30 ± 5 years, 180 ± 6 cm and 74 ± 13 kg, respectively, were enrolled for a randomized cross-over trial.The ethics committee of Copenhagen, Denmark approved the study (H-19013365) which was conducted in accordance with the Declaration of Helsinki.All subjects were informed orally and in writing about potential risks and discomforts associated with participation before written consent was obtained.
The participants completed three experimental trials in a randomized cross-over design (Figure 1). Figure 2 illustrates pictures of the applied equipment.The experimental trials were separated by at least 48 h and all three trials were completed within 3 weeks.Each trial included a single CO-rebreathing procedure on the first day followed by a second and third CO-rebreathing procedure conducted approximately 24 h later.The second and third CO-rebreathing procedures were separated by 5-10 min of rest.Each Figure 1.Schematic overview of the randomized trial protocol.the participants performed the co-rebreathing using three different devices (glass spirometer (red), semi-automated (green) and three-way valve (blue)).the first and second measurements were separated by 24 h while the second and third measurements were separated by 5-10 min.
experimental trial was conducted utilizing a different CO-rebreathing device to determine whether a difference in the measured total hemoglobin mass existed between the devices.The participants reported to the laboratory at least 2 h after any physical activity and were placed in a supine position for 10 min with the torso and legs slightly raised (~10°) before initiating the CO-rebreathing procedures.Following the last CO-rebreathing procedure, all participants were asked to rate the three devices in terms of comfort on a 1-10 comfort score scale with '1' being 'very uncomfortable' and '10' being 'very comfortable' .

Glass spirometer
The glass spirometer is a commonly used and commercially available device designed for CO-rebreathing and described in detail elsewhere [31].The participants equipped a nose clip and exhaled to residual lung volume followed by connection to the glass spirometer via a mouthpiece creating a closed system.Immediately upon connection, a valve was opened to a rebreathing bag containing 5 L of pure oxygen followed by administration of a bolus of ~1.0 (males) or ~0.8 (females) mL•kg body weight −1 chemically pure CO (99.997%,CO N47, Air Liquide, Paris, France) to the system.The CO was delivered by a 100-mL plastic syringe (Nemoto, Tokyo, Japan) which was filled with CO to the nearest 5 mL mark of the target dose.When preparing the CO-bolus, the syringe used for CO administration was flushed with CO before preparing the specific bolus to ensure that only CO was present in the syringe.Once the CO was administered, a timer was started, and the syringe was flushed three times with ~30 mL of the rebreathing gas mixture to ensure a complete delivery of CO to the system.Following two min of rebreathing, the participant exhaled to residual volume into the rebreathing bag and the valve here to be closed and the participant disconnected.A CO analyzer (Draeger, Lübeck, Germany) was used to evaluate whether a leak in the closed system occurred during the rebreathing period, to measure any leftover CO in the rebreathing bag and to measure end-tidal CO before and 3 min after the rebreathing period.At each experimental day, the rebreathing bag was inflated and left for 5 min to evaluate whether the system was airtight.

Semi-automated device
The second applied device was a semi-automated electromechanical CO-rebreathing device (Detalo Health, Birkerød, Denmark) covered by a stainless-steel casing.The device is controlled using a dedicated software on an external computer and connected to external CO and pure oxygen gas cylinders.The rebreathing is performed in an integrated rebreathing circuit, which at each experimental day was tested for leaks using the software's 'leak test' function.The rebreathing circuit was closed at the mouthpiece access, the 6-L rebreathing bag was inflated, and a stable rebreathing bag volume for 3-4 min indicated an airtight system.In addition, the internal chamber delivering CO to the rebreathing system was flushed with CO one to two times before measuring to secure only CO was present in the chamber.A small cylinder containing limestone was connected in series with the rebreathing system to absorb carbon dioxide and enable an extended rebreathing period.
Upon measurement, the participants equipped a nose clip and were connected to the rebreathing circuit via a mouthpiece which allowed access to room air.Next, by using the software, oxygen was supplied to the system followed by closing the rebreathing circuit and administration of CO molecules to the rebreathing circuit corresponding to 1.0 (males) or 0.8 (females) mL•kg body weight −1 chemically pure CO (99.997%,CO N47, Air Liquide, Paris, France) at a standard ambient temperature (25 °C) and pressure (760 mmHg) to the rebreathing circuit.The software added additional oxygen during rebreathing when the bag volume decreased to a certain extent.The software was set to a rebreathing period of 6 min, at which time the participants exhaled to residual volume, disconnected from the mouthpiece, which was rapidly closed with a stopper, and the rebreathing period was stopped within the software.A CO analyzer (Draeger, Lübeck, Germany) was used to measure any leftover CO in the rebreathing bag using an outlet on the device and to measure end-tidal CO before and 2 min after the rebreathing period.

Plastic three-way valve
The third device was a standard plastic three-way valve designed for pulmonary measurements, with a rotating valve allowing control of flow direction from one of the three exits to one of the two other exits like those commercially available https:// www.rudolphkc.com/product-page/three-way-y-shape-manual-stopcock-type).The three exits of the three-way valve were used for (1) mouthpiece connection, (2) the 5-L rebreathing bag containing pure oxygen and (3) an access to room air.The participants equipped a nose clip and connected to the three-way valve via a mouthpiece, which was set to give access to room air.The participants then exhaled to residual volume, and the three-way valve was turned to create a closed system between the participant and the rebreathing bag.Next, a bolus of ~1.0 (males) or ~0.8 (females) mL•kg body weight −1 chemically pure CO (99.997%,CO N47, Air Liquide, Paris, France) was administered to the system through a 100-mL plastic syringe (Nemoto, Tokyo, Japan) connected via a luer lock inlet positioned on the adapter connecting the rebreathing bag and the three-way valve.When preparing the CO-bolus, the syringe used for CO administration was flushed with CO before preparing the specific bolus to ensure that only CO was present in the syringe.The syringe was filled with CO to the nearest 5 mL mark of the target dose.A timer was started following CO administration, and the syringe was flushed three times with ~30 mL of the rebreathing gas mixture to ensure a complete delivery of CO the system.The participants exhaled to residual volume after 2 min, the valve was turned hereby closing off the rebreathing bag and provided the participant access to room air after which they were disconnected.A CO analyzer was used to evaluate leaks, leftover CO and end-tidal CO as described with the glass spirometer.At each experimental day, the rebreathing bag was inflated and left for 5 min to evaluate whether the system was airtight.

Blood sampling
Four baseline capillary samples were collected from a fingertip in 35-µL pre-heparinized tubes (Clinitubes, Radiometer, Brønshøj, Denmark) before and after the rebreathing period of 7 min (i.e. 9 min from the initiation) when using the glass spirometer and the three-way valve and 4 min after (i.e.10 min from initiation) when using the semi-automated device.All four blood samples were analyzed for carboxyhemoglobin percentage (%HbCO) on a blood gas analyzer (ABL800 Flex, Brønshøj, Radiometer, Denmark).All blood samples were collected from a hand pre-heated by a heating pad.

Modifications when performing duplicate measures immediately following one another
When two CO-rebreathing procedures were conducted immediately following one another, the timing of the baseline blood samples was slightly modified for the second measurement.Specifically, the participants were connected to the closed rebreathing circuit containing pure oxygen before baseline blood samples were collected.Immediately upon connection to the closed circuit, the baseline blood samples were collected as fast as possible and typically within 30s.Had the standard protocol been applied, i.e. collection of baseline samples before connection to the closed rebreathing circuit, the participants would have lost CO to the atmosphere via exhalation in the time between blood sampling and connection to the closed circuit.Such loss is expected to be high due to the CO exposure from the first CO-rebreathing procedure, but the loss is remedied by the protocol adjustment.Following blood sample collection, the CO was administered in identical manner as for the first measurement and a timer was started.The adjustment was similar for all devices and the remaining protocol was identical to that of the first CO-rebreathing procedure for each device The CO-rebreathing devices were thoroughly flushed with room air between measurements to secure removal of excess CO.

Calculations
The total hemoglobin mass independent of CO-rebreathing device was calculated based on the principles described elsewhere [32] using Eq.(I), where K, MCO and Δ%HbCO was calculated as in Eqs.(II), (III) and (IV), respectively.The constant of 1.39 represents Hüfners number for the CO binding capacity of hemoglobin.I. Total hemoglobin mass (g) = K × MCO × 100/ (Δ%HbCO × 1.39) II.K = (ambient barometric pressure mmHg × 273°K)/ (760 mmHg × ambient temperature °K) III.MCO = CO volume administered to the system -CO volume not bound to hemoglobin IV.Δ%HbCO = %HbCO post -%HbCO baseline The applied CO volume not bound to hemoglobin was calculated as the sum of CO leftover in the rebreathing system, CO lost via exhalation and CO lost to myoglobin.Leftover CO in the rebreathing system was calculated as the CO concentration within the rebreathing bag multiplied by sum of the rebreathing bag volume (5 L for glass spirometer and three-way valve, whereas a measured volume is provided by the semi-automated device) and the estimated residual lung volume (1.5 L for males, 1.2 L for females).The loss of CO via exhalation was calculated as the difference in end-tidal CO concentration from before to after the CO-rebreathing multiplied by a standard alveolar ventilation of 5 L and multiplied by the number of minutes between disconnection and blood sampling.Finally, the loss of CO to myoglobin was calculated as 0.3% per minute [33] multiplied by the time from CO exposure to blood sampling.
The software of the semi-automated device applies a different set of calculations to estimate total hemoglobin mass based on the number of CO molecules rather than the volume [34] and according to the device guidelines, loss of CO to myoglobin or via exhalation, is not taken into account.To enable a comparison between all three devices, the end-tidal CO was determined before and after the rebreathing period for all three procedures.Furthermore, the number of administered CO molecules (nCO), which is disclosed by the software of the semi-automated device, was converted to a volume of CO using the ideal gas law (equation (V)) where R is the gas constant (0.08206 L × atm/mol × K), T is the temperature in Kelvin, and P is the atmospheric pressure.

V. CO volume administered = (nCO × R × T)/P
The calculated administered CO volume was then used in equation (I)-(IV) for calculation of total hemoglobin mass including corrections for loss of CO to myoglobin and by exhalation allowing for a direct comparison between all three devices.

Statistics
A mixed linear model for repeated measures was applied.To assess whether a difference existed between the first and second measure of the 222 duplicates measures performed immediately following one another, a fixed factor of 'measure' (i.e.first or second measure) was included.In addition, 'device' (i.e.glass spirometer, three-way valve, and semi-automated device) and 'device × measure' were included as fixed factors to assess whether a difference between devices existed.
In the randomized trial, a similar mixed linear model for repeated measures was applied.To assess whether a similar total hemoglobin mass was obtained by duplicate measures interspersed by 24 h and 5-10 min, the average total hemoglobin mass of the first and second measurement was compared to the average of the second and third measurement with inclusion of 'time' (i.e.24 h or 5-10 min) as a fixed factor.In addition, a fixed factor of 'device' (i.e.glass spirometer, three-way valve, and semi-automated device) and 'time × device' were included to assess whether a difference between devices existed.To determine if the participants comfort scores were different between devices a linear mixed model with a fixed factor of 'method' was applied.Finally, a linear regression analysis was used to determine the association between the determined hemoglobin masses obtained from the three devices.
Repeated measures over time were considered by including a random effect of participants.If a fixed main effect or interaction was significant, a post hoc analysis was performed using a Sidâk adjusted pairwise comparison.SPSS version 28.0.0 (IBM SPSS, Chicago, USA) was used for statistical analyses and figures were made using GraphPad Prism 10.1.0(GraphPad Software; USA) or PowerPoint (Microsoft Office, Redmond, USA).The results are presented as means ± standard deviation (SD) and the level of significance was set at p < 0.05.The typical error or total hemoglobin mass measurements were calculated as described elsewhere [23].

Immediate reproducibility
When analyzing the 222 CO-rebreathing measures performed immediately following one another, an effect of 'measure' existed (p < 0.05), revealing that the total hemoglobin mass of 852 ± 195 g at the first measurement was higher (p < 0.05) than the total hemoglobin mass of 849 ± 192 g obtained at the second measure.In addition, a significant 'measure × device' interaction was evident (p < 0.001).The pairwise analysis revealed that the total hemoglobin mass obtained by the semi-automated device and three-way valve was 13 ± 23 g lower (p < 0.001) and 3 ± 19 g higher (p < 0.05), respectively, at the second measure.There was no statistical difference between the first and second measure for the glass-spirometer although the second measure was numerically 6 ± 23 g lower.
The overall typical error was 1.1%, ranging from 1.0% to 1.2% within devices.A Bland-Altman plot illustrating the reproducibility of the duplicate measures is presented in Figure 3.The HbCO was 0.9 ± 0.2% and 6.0 ± 0.7% before and after the first measurement, and 5.9 ± 0.7% and Figure 3. bland-altman plots for individual total hemoglobin mass differences between two co-rebreathing measures performed immediately following one another using the glass spirometer (red circles), the semi-automated device (green circles) and the three-way valve (blue circles).the solid line indicates the mean difference while the dotted lines indicate the 95% limits of agreement (loa), which was calculated using all datapoints.
10.9 ± 1.3% before and after the second measurement, respectively.The individual HbCO after the second measurement ranged from 7.9% to 16.5%.

Randomized trial
There was no significant effect of 'time' or 'time × device' in the randomized trial.However, a significant effect of the applied device existed (p < 0.001).Specifically, the total hemoglobin mass obtained by the semi-automated device was 53 ± 30 g and 47 ± 31 g higher (p < 0.001) than that obtained by the glass spirometer, and 56 ± 38 g and 48 ± 34 g higher (p < 0.001) than that obtained by the three-way valve when duplicate measures were separated by 24 h and 5-10 min, respectively.The total hemoglobin mass obtained by the glass spirometer and the three-way valve was not different at any time point.Furthermore, an almost perfect correlation between total hemoglobin mass obtained with the three-way valve (Y = 1.012 × X-11.15;R 2 =0.98, p < 0.001) and the semi-automated device (Y = 1.057 × X-3.00; R 2 =0.97, p < 0.001) compared to the glass-spirometer existed (see supplementary Figure 1).
The typical error was 0.9% and 1.0% for the glass-spirometer, 1.0% and 1.2% for the three-way valve and 0.7% and 1.3% for the semi-automated device when duplicate measured were separated by 24 h or 5-10 min, respectively.Bland-Altman plots of the reproducibility are presented in Figure 3 and the individual total hemoglobin mass for the two duplicate measurement for each device is reported in Table 1.The HbCO obtained in the randomized trial are available in Table 2.
No differences were found for the comfort scores, which were 8.7 ± 2.0, 7.3 ± 1.1, and 8.2 ± 1.5 for the plastic valve, the glass-spirometer, and the semi-automated device, respectively.

Effect of CO loss correction
Importantly, the outcomes for device comparisons and reproducibility did not change when total hemoglobin masses were left uncorrected for a loss of CO to myoglobin and via exhalation.
However, the difference in total hemoglobin mass was up to 98 ± 37 g (p < 0.001) when comparing the total hemoglobin mass of the semi-automated device left uncorrected for CO-loss, and the CO-loss corrected total hemoglobin mass by the glass-spirometer.See Figures 4 and 5 for the remaining comparisons.In addition, it should be noted that when calculating total hemoglobin mass using either the applied CO volume and Hüfners number [31] as in the present study or the applied number of CO molecules and the molar weight of hemoglobin [34], similar results were obtained (≤2 g difference) when the same corrections were applied.

Discussion
The major finding of the present study was that duplicate measures of total hemoglobin mass by CO-rebreathing conducted immediately following one another was feasible, reliable with a typical error of ~1% and provide a comparable total hemoglobin mass to duplicate measures conducted one day apart.In addition, the total hemoglobin mass was similar when measured by a custom-designed glass-spirometer or a standard three-way plastic valve designed for pulmonary measurements, whereas the total hemoglobin mass was ~6% (~50-60 g) higher when measured by a semi-automated CO-rebreathing device.Importantly, differences up to ~13% (~100 g) were evident between devices depending on whether a correction for loss of CO to myoglobin and via exhalation was included, demonstrating the importance of accurately reporting the applied methodology and considering methodological differences when comparing hemoglobin values across studies.
The HbCO reached ~10.8% in the present study following immediate duplicate measures (Table 2) and ranged from 9.2% to 13.1% within individuals, which reflects the wide range of 6.5-16.0g hemoglobin•kg body weight −1 .Previous studies performing duplicate CO-rebreathing procedures at the same day aimed at maintaining HbCO below 10%, which with physical activity and hyperoxia for increased HbCO clearance between the two measures required a time gap of 2 h [24,25].However, increasing HbCO to ~18% by CO-rebreathing in healthy males only induced a light headache in <50% of the individuals [5], which align with the suggested upper limit of 15% for HbCO to avoid side-effects [35].Moreover, HbCO levels >35% are required to affect a cognitive task such as driving [36] and a dose of >1000 mL CO is required for an adult male to induce life-threatening symptoms [37].In the present paper, commonly observed side-effects of CO-intoxication, e.g.dizziness, nausea or lethargy [38], was not systematically quantified but two participants expressed a light headache after the immediate duplicate measurements performed with the three-way valve.The present study suggests that immediate duplicate measures in healthy individuals is feasible and does not compromise safety with respect to HbCO accumulation when a dose of 1.0 (males) or 0.8 (females) mL•kg body weight −1 CO is applied.
A minor difference between the first and second measurement of 13 g was evident for the semi-automated device (n = 82) and 3 g for the three-way valve (n = 130), whereas no statistical difference was evident for the glass-spirometer.The latter is expectedly due to the lower statistical power (n = 10), as a numeric difference of ~6 g existed.The systematic difference between the first and second measurement of 3-13 g is comparable to the random measurement error of 1-2% and appears negligible for detection of physiologically relevant changes in total hemoglobin mass.
Collectively, the present study demonstrates that performing two CO-rebreathing procedures immediately following one another is a feasible and reliable strategy for overcoming the logistical challenge of duplicate measures of total hemoglobin mass.

Device comparison
Despite a nearly perfect correlation between devices (R 2 =0.97-98), a ~6% higher total hemoglobin mass, corresponding to ~50-60 g, was observed in measurements performed by the semi-automated device in comparison to the two other devices.This correspond to the ~7% higher total hemoglobin mass observed by others for the semi-automated device compared to the glass-spirometer when both methods were similarly corrected for loss of CO [29].The higher total hemoglobin mass was suggested to partly being caused by differences in body posture.However, in the present study all procedures were performed in supine position, indicating that body posture is not responsible for the differences observed between devices.This is supported by a recent study demonstrating that seated and supine CO-rebreathing procedures provide a similar total hemoglobin mass when performed with the glass-spirometer [30].
Another potential explanation is the different length of the rebreathing protocol.However, a comparison between CO-rebreathing for 2 and 10 min using the semi-automated device provided a similar total hemoglobin mass [28].Additionally, studies utilizing other devices for comparison between long (≥10 min) and short (2-min) rebreathing periods find a similar total hemoglobin mass when corrections for CO-loss are similar [31,32,39], whereas others find a higher [40] or lower [41] total hemoglobin mass of the long rebreathing procedure.As such, it does not appear that the rebreathing duration can explain the difference.
It is also important to consider the timing of blood sampling, which differed marginally from minute 9 for the glass-spirometer and three-way valve to minute 10 for the semi-automated device, after initiating CO-rebreathing.Nevertheless, as the mixing of CO is expected to be complete within 4-6 min in the supine position [5,30] and the loss of CO was accounted for, sample timing does not appear to explain the discrepancy either.Moreover, the randomized cross-over trial was conducted in a period of up to 3 weeks for each individual, which is unlikely to induce biological variation at a magnitude of the observed differences [42] and any variation must be expected to be randomly distributed across devices.
Furthermore, the accuracy of the syringes used for manual CO administration was retrospectively evaluated as a possible explanation for the discrepancy.The syringe was filled with water to a target dose, and the water was ejected onto a scale.The ejected water amount differed less than 1 mL from the target dose for all syringes.In addition, as the present study flushed the syringe three times with ~30 mL of the rebreathing gas mixture to ensure a complete delivery of CO to the system, any CO within the dead space of the syringe would also be administered, which was quantified to ~0.5 mL.However, administering 1 mL of CO above the target dose only accounts for ~10 g in hemoglobin mass, and cannot explain the observed discrepancy of 50-60 g between the manual and semi-automated setup.
Notably, in healthy adults a manual 10-min CO-rebreathing procedure measures a ~250 mL higher red blood cell volume in comparison to the per convention clinical state-of-the-art standard method for determination of intravascular volumes, i.e. a dual-infusion procedure utilizing sodium pertechnetate ( 99m Tc) labeled red blood cells and iodine-125 labeled human serum albumin [14].Interestingly, the semi-automated device measures an additional ~200 mL higher, i.e. ~450 mL higher red blood cell volume when compared to the dual-infusion procedure [28,43].Curiously, in the present study the 50 g higher hemoglobin mass of the semi-automated device in comparison to the manual procedures corresponded to a ~150 mL red blood cell volume when using the participants own hematocrit and [Hb] for calculation (data not shown).Although these studies corroborate the current findings that the semi-automated CO-rebreathing device determines a higher total hemoglobin mass when compared to manual CO-rebreathing approaches, it is important to emphasize that the current findings do not indicate which method provides the most valid determination.
The comparison between devices has been a recent topic of debate for the developers of the commercially available ones [44,45], but they do not provide data explaining the discrepancy.Accordingly, it appears that when the same set of assumptions and calculations are used the semi-automated device provides a 6-7% higher total hemoglobin mass when compared to the glass-spirometer and three-way valve for reasons unknown.

Three-way valve
In addition, the current study also demonstrates that the use of a standard three-way valve designed for pulmonary measurements is a reliable device for measuring total hemoglobin mass and provides results and comfort on par with commercially available devices.As the plastic valve has the advantage of being less fragile than glass and cheaper than the semi-automated device, it appears to be a viable alternative for CO-rebreathing.The plastic valve also enables access to room air prior to initiating the rebreathing procedure while the user is fully mounted to the device, including nose clip, which contrasts with the glass spirometer.Although not quantified, it is our experience that most of the errors occurring during the rebreathing (e.g.user leaks air from the closed system) takes place when the rebreathing is initiated, which in our hands appear to be remedied by the three-way valve system.

Considerations when comparing results across studies
Finally, it is important to emphasize the importance of reporting and considering whether a loss of CO to myoglobin and via exhalation is included in the calculation of total hemoglobin mass.When the device-specific recommendations for correction of CO-loss are followed, the difference between devices was amplified from 6% to 13%, which in this study amounted to ~100 g, as the semi-automated device per default does not correct for a loss of CO.Such a difference is of clear physiological relevance.The importance of how total hemoglobin mass is calculated has been reported previously [39] and is reinforced by the present study.Consequently, it is paramount to report the details of how total hemoglobin mass is calculated to allow for a potential comparison across studies.

Conclusion
The present study demonstrates that it is feasible and reliable to perform two CO-rebreathing procedures immediately following one another with a typical error of ~1% when doses of 0.8-1.0mL CO•kg body weight −1 mass are applied.In addition, the total hemoglobin mass was similar between the custom-designed glass-spirometer and the standard three-way plastic valve, whereas the semi-automated CO-rebreathing device estimated a ~6% higher total hemoglobin mass for unknown reasons.However, a difference of ~13% exists if the device-specific recommendations of CO-loss corrections are followed.Future studies should attempt to decipher where this difference origins from.

Figure 2 .
Figure 2. Pictures of the three devices investigated.Picture a: the glass-spirometer, picture b: the semi-automated device overview, and picture c: the three-way plastic valve.

Figure 4 .
Figure 4. bland-altman plots for individual total hemoglobin mass differences obtained in the randomized trial.the left panel represent the first and second measurement (separated by 24 h) and the right panel the second and third measurement (separated by 5-10 min) using the glass spirometer (red circles), the semi-automated device (green circles) and the three-way valve (blue circles).the solid line indicates the mean difference while the dotted lines indicate the 95% limits of agreement (loa).

Figure 5 .
Figure 5. total hemoglobin mass measured by the three different devices in the randomized trial.each column represents the average of all three measurements in the randomized trial with one standard deviation as error bars.the '− co loss correction' indicate that the calculation of total hemoglobin mass was not corrected for loss of co to myoglobin and via exhalation, whereas the '+co loss correction' indicate that it was.Statistically significant differences are as follows (all p < 0.001): 'a' and 'd' when compared with glass-spirometer without (−) and with (+) co loss correction, respectively.'b' and 'e' compared with semi-automated device without (−) and with (+) co loss, respectively.'c' compared with three-way valve without (−) co loss correction.

Table 1 .
the individual average total hemoglobin mass in grams obtained in the randomized trial by each of the three devices in duplicates measures separated by 24 h (measurement 1 and 2) or 5-10 min (immediate, measurement 2 and 3).