Pre-analytical validity of arterial blood gas samples: A prospective experimental study on changes derived from time delay and mechanical stress

doi:10.21203/rs.3.rs-4319836/v1

Background: Time delays and mechanical stress of samples obtained for Point of Care (POC) blood gas analyses are common; however, their influence on the results of these analyses has not been systematically investigated. Our study aimed to investigate the effect of prolonged time before analysis and mechanical manipulation on pre-analytical stability of biomarkers and thus the validity of the results of blood gas analyses.

Methods: We collected blood samples from 240 patients in a university surgical intensive care unit. These samples were immediately analyzed following the clinical standard operating procedures. Subsequently, the sample containers were allowed to rest for 60 min, then subjected to standardized mechanical forces, and analyzed again. We analyzed 13 typical blood gas biomarkers, comprising respiratory gases, electrolytes, and protein biomarkers. Bland–Altman plots were prepared to analyze the differences between the test runs. The differences between the test groups were compared against the official limits of accuracy specified in the German requirements for quality assurance of medical laboratory tests.

Results: For hemoglobin, creatinine, glucose, and electrolytes (including calcium, sodium, chlorine, and bicarbonate), the agreement between the immediate and post-interference-treatment analyses was within the ranges specified in the official requirements. For pH and potassium, the deviations were outside the quality assurance ranges but within a clinically acceptable measurement accuracy. Only oxygen partial pressure and lactate levels were altered to such an extent that they can no longer be used for clinical purposes.

Conclusion: Even after a 60 minutes time delay and excessive mechanical stress, selected blood gas analysis biomarkers such as Hemoglobin, Glucose, Sodium, Calcium, Chloride, and Bicarbonate could be considered valid. Potassium and pCO2 were altered but suitable for approximation purposes. Findings for pO2 and Lactate were generally incorrect. In the future, in selected settings, these findings can aid in reducing unnecessary blood sampling in vulnerable patients.

Blood

gas

analysis

time delay

mechanical stress

intensive care unit

biomarkers

oxygen partial pressure

In anesthesia and intensive care medicine, blood gas analysis is a basic diagnostic tool for at-risk and critically ill patients. Careful handling of the samples after collection and immediate analysis are accepted standards. Nevertheless, numerous situations occur in everyday clinical practice that could lead to unintentional delays in the analysis or external interference with samples. These situations could include lengthy transport routes and equipment downtime owing to maintenance, human negligence, or mechanical interference caused by mishandling. The current literature does not answer whether samples remain suitable for analysis after prolonged periods and/or mechanical interference. This uncertainty often prompts clinicians to obtain new blood samples, incurring costs, time, and potential patient discomfort and harm associated with renewed blood loss or repeated vascular punctures [1, 2].

Existing literature has only focused on certain aspects of the pre-analytical stability of blood gas samples. For example, Knowles and Harsten detected changes in partial pressures of respiratory gases (pO₂ and pCO₂) 30 and 60 min post-collection. Moreover, Smeenk et al. [3] described prolonged stability of pO₂ when using glass syringes or ice-water storage. However, beyond respiratory gases, modern blood gas analyzers measure various clinically relevant analytes, of ionic or organic-small-molecular character (for example, c(sodium), c(potassium), c(calcium), c(glucose), and c(lactate)). Existing data on the possible pre-analytical influence of these numerous parameters are limited. In a recent study, we demonstrated that time delays and mechanical stress on venous blood samples in an out-of-hospital emergency medical service setting did not lead to any clinically relevant alterations in various electrolytes, small organic molecules, or protein biomarkers [4].

Therefore, this study aimed to investigate the effect of prolonged time before analysis and mechanical manipulation on pre-analytical stability and the validity of the results of blood gas analyses.

We aimed to answer the following questions:

First, whether changes are systematic or erratic, statistically significant, or clinically relevant. Second, whether in terms of data quality, results derived from delayed analysis or samples subjected to mechanical stress remain suitable for clinical purposes and decision-making under pertinent clinical circumstances.

Aim, design, and setting

This study aimed to investigate the effect of prolonged time before analysis and mechanical manipulation on pre-analytical stability and the validity of the results of blood gas analyses. The Ethics Committee of the Friedrich-Alexander University, Erlangen-Nuremberg, Germany, reviewed and approved the study proposal (vote 22-124-B). Prospective participants or their legal representatives provided both verbal and written consent in advance and participated voluntarily in the study. The study population comprised patients from all surgical specialties who were treated at our interdisciplinary tertiary-level intensive care unit. Pregnant women and minors were excluded from participation.

Blood samples

The procedure for collecting blood samples was standardized, ensuring compliance with the syringe manufacturer's specified product manuals and in-house standard operating procedures regarding hygiene and workplace safety.

Blood samples were collected from arterial catheters, using a "safe Pico Aspirator" from Radiometer Medical ApS, Brønshøj, Denmark. Subsequently, the samples were analyzed using blood gas analyzers (ABLflex800) (Radiometer Medical ApS, Brønshøj, Denmark) located within the intensive care unit.

Table 1 provides a list of the measured parameters, including the type of analyzer, analytical methods, and legally required accuracy of analysis. Parameters are divided into two groups: those directly measured from the sample, such as pH, pCO₂, pO₂, ctHb, cK⁺, cNa, cCl, cCa²⁺, cGluc, cCrea, and cLac, and those derived via calculation, such as HCO₃ and SO₂. The case number was not estimated in advance. Instead, it was derived based on local resources and general practicability.

Table 1

Biomarkers and analyzers, analytical methods, and quality criteria.
Biomarkers	Analyzer model	Manufacturer	Test method	Frequency of quality controls (three shifts per day)	Acceptable Root Mean Standard Deviation as defined by Rili-Beak	Range of intrest as defined by Rili- Baek	Unit
Calcium	ABL800 Flex	Radiometer Medical ApS, Brønshøj, Denmark	Potentiometric method	Once per 8 h shift	14% 7.5%	0.2 to ≤ 1 > 1–2.5	mmol/L
Chlorides	ABL800 Flex	Radiometer Medical ApS, Brønshøj, Denmark	Potentiometric method	Once per 8 h shift	4.5%	70–150	mmol/L
Creatinine	ABL800 Flex	Radiometer Medical ApS, Brønshøj, Denmark	Potentiometric method	Once per 8 h shift	11.5	0.5–10	mg/L
Glucose	ABL800 Flex	Radiometer Medical ApS, Brønshøj, Denmark	Potentiometric method	Once per 8 h shift	11%	40–400 2.2–22	mg/dL mmol/L
Hemoglobin	ABL800 Flex	Radiometer Medical ApS, Brønshøj, Denmark	Potentiometric method	Once per 8 h shift	4.0%	2–20	g/dL
Potassium	ABL800 Flex	Radiometer Medical ApS, Brønshøj, Denmark	Potentiometric method	Once per 8 h shift	4.5%	2–8	mmol/L
Lactate	ABL800 Flex	Radiometer Medical ApS, Brønshøj, Denmark	Potentiometric method	Once per 8 h shift	11.0%	1–10	mmol/L
Sodium	ABL800 Flex	Radiometer Medical ApS, Brønshøj, Denmark	Potentiometric method	Once per 8 h shift	3.0%	110–180	mmol/L
pCO₂	ABL800 Flex	Radiometer Medical ApS, Brønshøj, Denmark	Potentiometric method	Once per 8 h shift	7.5% 6.5%	≤ 35 > 35	mmHg
pH	ABL800 Flex	Radiometer Medical ApS, Brønshøj, Denmark	Potentiometric method	Once per 8 h shift	0.4%	6.75–7.80
pO₂	ABL800 Flex	Radiometer Medical ApS, Brønshøj, Denmark	Potentiometric method	Once per 8 h shift	5.5% 7.0% 11.0%	> 125–350 > 80 to ≤ 125 40 to ≤ 80	mmHg
HCO₃	ABL800 Flex	Radiometer Medical ApS, Brønshøj, Denmark	calculated
SO₂	ABL800 Flex	Radiometer Medical ApS, Brønshøj, Denmark	calculated
For further detailed information on laboratory medical examination methods, we recommend consulting the manufacturer's user manual and the quality criteria prescribed by the German Medical Association available on the Rili-Baek homepage (https://www.bundesaerztekammer.de/themen/aerzte/qualitaetssicherung/richtlinien-leitlinien-empfehlungen-stellungnahmen)

Test scenario

Blood samples were sequentially tested for the first time immediately after collection and for the second time 60 min post-mechanical stress. First, blood samples were collected as a part of routine clinical practice. A first blood gas analysis was performed immediately after the samples were obtained [5]; the remaining quantity of blood was not discarded but preserved. For this purpose, the sample containers were emptied of residual air and stored on a tray next to the blood gas analyzer at room temperature during the waiting period between the first and second analyses. They were not shielded from sunlight. Before the second analysis, the tubes were subjected to a standardized mechanical stress scenario. Each sample was dropped 10 times from a height of 85 cm (table height) and shaken vigorously 10 more times. The same samples were analyzed the second time to compare the paired results 60 min after the first analysis. The results of the second series of measurements were used only for this study and not for any clinical decisions.

Data analysis and statistics

Bland–Altman plots were used to compare the combined effects of mechanical stress and time delay on arterial blood gas samples [6]. Values from blood samples that were directly analyzed were compared with those analyzed after 60 min delay and mechanical manipulation.

Every Bland–Altman plot containes several lines. The central solid line in the graph represents the mean values of the differences in measured parameters. The two dashed lines above and below the solid line represent the upper and lower limits of agreement, respectively, which correspond to a maximum 1.96-fold deviation from the mean value upward and downwards. Typically, 95% of the measured values were within these limits.

As an external criterion for determining negligible differences, we used the specified validity ranges of the respective biomarkers following the German Medical Association guidelines for the quality assurance of laboratory medical examinations (Rili-Baek) [7].

Rili-Baek specifies the basic requirements for the quality management and assurance of laboratory medical examinations in Germany. Among other factors, the validity of individual test results within their respective validity ranges is defined. Hence, we assumed that the results can be regarded as equivalent and deviations negligible if they lie within the error margins of the Rili-Baek [7]. The validity ranges are shown in the diagrams by spreading green lines. The analysis was performed using SPSS and Microsoft Excel (IBM SPSS Statistics Version: 28.0.1.1(15)/ Microsoft® Excel® 2019 MSO (16.0.14332.20563) 32-Bit).

In total, 240 blood samples were collected. Thirteen biomarkers were analyzed, 11 of which were directly measured using the device, and two were derived via calculation. No missing data were reported. Each analyzed parameter was investigated for differences between the measurements using Bland–Altman plots. Figures 1–14 present the graphs for each parameter. Table 2 provides an overview of all the parameters, standard deviations, and confidence intervals.

Table 2

Overview of analyzed biomarkers
Biomarkers in alphabetical order	Unit	Number of measurements	Mean of differences	CI Mean	Upper limit of agreement	CI Upper limit of agreement	Lower limit of agreement	CI lower limit of agreement
Calcium	mmol/l	240	-0.0073	-0.0074	0.035	-0.0045	-0.049	-0.010
Chlorides	mmol/l	240	0.104	0.1019	2.836	0.281	-2.628	-0.073
Creatinine	mg/L	239	-0.002	-0.0054	0.162	0.009	-0.166	-0.013
Glucose	mg/L	240	-4.675	-4.8704	4.401	-4.086	-13.751	-5.263
Hemoglobin	g/dL	240	0.098	0.0806	0.518	0.1260	-0.321	0.072
Potassium	mmol/L	240	0.2189	0.2067	0.619	0.245	-0.181	0.193
Lactate	mmol/L	240	0.609	0.6047	1.074	0.639	0.145	0.579
Sodium	mmol/L	240	-0.077	-0.963	2.465	0.088	-2.619	-0.242
pCO₂	mmHg	240	1.196	1.2208	1.499	1.372	-3.893	1.022
pH		240	-0.017	-0.0171	0.044	-0.015	-0.009	-0.019
pO₂	mmol/L	240	33.740	33.7935	80.456	36.771	-12.976	30.709
HCO₃	mmol/L	240	-0.373	-0.4005	0.748	-0.300	-1.494	-0.4461
SO₂	%	240	1.855	1.7199	6.184	2.136	-2.475	1.577

We discovered that the deviations in the results of all parameters except K+, lactate, pCO₂, and pO₂ were within the specified limits of Rili-BaeK and the standard deviation of the Bland-Altman plots.

The measured values for K+, lactate, PCO₂, and PO₂ indicated that deviations in the results were within the standard deviation of the Bland–Altman plot. However, we observed a substantial deviation from the validity criteria specified by Rili-Baek. Even a 10% deviation corridor of the measured values from the original values, which we introduced based on possible clinical decision-making strategies, could often not be met.

For potassium, 52.5% (126/240) of the values fell within the limits specified by Rili-Baek. Additionally, 88.3% (212/240) of the measured results were within the proposed 10% margin of deviation. Furthermore, 93.75% (225/240) of measurements showed an overall increase in potassium concentration.

The measured values for lactate were rarely (1/240: 0.4%) within the validity range specified by Rili-Baek. Similarly, only 1 out of 240 values fell within an additional 10% corridor. Moreover, 99.6% (239/240) of the measured values showed an increase in lactate concentration.

Regarding the measured values of pCO₂, we discovered that 80.4% (202/240) were within the required measurement ranges of the Rili-Baek. Moreover, 97.9% (235/240) of the values fell within the 10% corridor. In total, 85.4% (205/240) showed an overall increase in partial pressure.

The results for the measured pO₂ values revealed that 5.4% (13/240) were within the Rili-Baek limits, and 7.9% (19/240) were within the additional 10% corridor. In total, 95% (228/240) showed an overall increase in partial pressure.

The results for the measured pH values showed that 84.5% (203/240) were within the Rili-Baek limits, with 100% (240/240) falling within the additional 10% corridor. In total, 81.6% (196/240) showed an overall decrease in the measured value.

The results for the measured glucose values indicated that 98.75% (237/240) were within the margin of error specified by the Rili-Baek. Furthermore, 89.2% (214/240) showed a decrease in the measured value.

Careful handling and immediate analysis of blood samples are medical standards. However, in the bustling and stressful environment of an intensive care unit, blood samples might unintentionally undergo rule violations. The novel findings of this study demonstrate that many biomarkers from an arterial blood gas sample are robust against pre-analytical mechanical stress or time delay before analysis.

To that end, we compared the test results of an arterial blood gas analysis (BGA) after immediate analysis with those from the same sample after a combination of a defined time delay and simulated mechanical stress. We deliberately exaggerated both confounding variables (time and mechanical forces) to a level greater than expected under real-life conditions. The rationale for this experimental approach was that biomarkers proving stable against supramaximal scenarios would certainly also remain stable against stressors in everyday circumstances.

In our study, the specifications for technical quality control of the analyzers, in accordance with Rili-Baek, were used to define what would constitute a negligible deviation. Deviations of the measured values that were within the methodological measurement inaccuracy were regarded as clinically equivalent. Therefore, a subjective interpretation of values, prone to bias, was initially unnecessary.

The most common blood gas analysis parameters were included in the dataset. The results are interpreted in three different ways.

1. Dissolved gases undergo relevant diffusion processes across the walls of the sample containers. Based on the partial pressure differences between the blood and room air, a net inflow of oxygen from the room air into the sample and a net outflow of carbon dioxide from the sample into the room air occurs [8/9].

In most cases, the increase in pO₂ exceeded the margin specified in Rili–Baek. Substantial oxygen enrichment of the samples (mean + 38 mmHg) without any meaningful relation to the patient's condition was observed. Therefore, an interpretation of the oxygenation performance of the patients was categorically impossible in our test scenario.

For CO₂, we observed a slight average increase in the partial pressure in the sample vessel (mean +1.2 mmHg; upper limit of agreement < 4 mmHg). At first glance, this observation appears to contradict the net CO₂ loss from the containers we described earlier. However, as a result of the substantial increase in hemoglobin oxygenation in the sample vessel, the CO₂ previously bound to hemoglobin was released. This so-called Haldane effect obviously outweighs the diffusion loss during the investigation period, and the sum of these effects results in a slight increase in the pCO₂of the sample [10].

These changes in pCO₂ were mostly within a 10% deviation margin that we set suggestively. We believe that such a +/- 10% corridor often is part of clinical decision-making patterns. As a consequence, a discussion is warranted whether the observed moderate changes in pCO₂ values should be viewed as numerically different but clinically similar and could therefore be used as an approximation of decarboxylation.

2. Some biomarkers of the erythrocyte metabolism (mean lactate +0.6 mmol/L; mean potassium +0.2 mmol/L) were also subject to changes outside the margin of error specified in the Rili-BAEK. While changes in lactate were far beyond a clinically acceptable margin of error, the majority of potassium measurements lay within the above-mentioned 10% corridor. Hence, we would like to spark a discussion, whether clinicians would still view these changes as an acceptable inaccuracy and should utilize those potassium values for therapeutic decisions.

3. Hemoglobin, creatinine, glucose, and electrolytes that are not closely related to erythrocyte metabolism are within the specified accuracy limits except for a few individual observations. Therefore, the results can be regarded as equivalent, justifying the use of these parameters for clinical decision-making, even after time delay and mechanical disturbance.

In summary, it is impossible to draw a uniform picture regarding the pre-analytical validity of our ill-treated blood samples. A differentiated interpretation of each biomarker based on its biochemical and physicochemical properties is required.

Naturally, our study has some strengths and weaknesses. One strength of our interpretative approach lies in using the purely technically derived error margins of the Rili-Baek as a basis for deciding whether value deviations should be regarded as acceptable or un-acceptable. Parameters that were within this predefined margin of error, even after our intervention, can be regarded as reliable with a high degree of probability and can be used for clinical decision-making.

However, some parameters fell into a gray area outside the interpretation certainty of the Rili-Baek margins but they remained within a clinically justified 10% error corridor. Consequently, the interpretation of these values remains debatable.

One limitation of this study is the incomplete selection of only a few standardized confounding factors. Other environmental factors that were not part of our experimental setup, such as temperature fluctuations and ultraviolet radiation, might also affect the pre-analytical stability of blood samples to an unknown extent.

Finally, it might be possible that the excessive degree of rule violations in sample treatment at the supra-everyday level was responsible for the large deviations in results for some biomarkers. One has to wonder, whether a milder form of treatment (for example, only a 30-min delay, only one drop) might have led to smaller changes, closer to Rili-Baek and thus to results that would have been clearer to interpret.

In our study, we demonstrated that time delay and mechanical stress before analysis of arterial blood gases had no relevant influence on the results for protein biomarkers (Hemoglobin, Creatinine), Glucose and certain electrolytes (Sodium, Calcium, Chloride, HCO3). pCO2 and Potassium are altered, but results might be considered for approximation purposes. Measurements of oxygenation (pO2) and lactate levels are generally incorrect after introducing time delay and mechanical stress.

Based on our data, it is not necessary for selected clinical questions to repeat a BGA if there was time delay or mechanical irritation of the sample before analysis. Despite its limitations, our study makes a valuable contribution to reducing unnecessary blood sampling in patients in the intensive care unit.

Future follow-up studies should consider the differential effects of time, mechanical force, or other factors on the pre-analytical stability of blood samples.

Rili-BAEK, Guidelines of the German Medical Association for Quality Assurance in Medical Laboratory Testing

Ethics approval and consent to participate

The Ethics Committee of Friedrich-Alexander-University, Erlangen-Nuremberg, Germany, approved this study under reference Number 22-124-B. Written informed consent was obtained from all participants before the experiments.

Consent for publication

Not applicable

Availability of data and materials

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests

The authors declare that they have no competing interests.

Funding

Not applicable.

Authors' contributions
MG conducted the experiments and performed statistical analyses. The present work was performed in fulfillment of the requirements for MG to obtain the academic degree "Dr. med." HI performed statistical analysis and critically revised the manuscript for intellectual content. FK and AM coordinated the study and critically revised the intellectual content of the manuscript. JP conceived the study and performed statistical analyses. All authors have read and approved the final manuscript.

Acknowledgments

We thank all volunteers for their contributions to our study.

Author’s information

1 Anesthesia group practice Bernard and Weber, 91054 Erlangen, Germany

2 Department of Anesthesiology, Erlangen University Hospital, 91054 Erlangen, Germany.

3 Faculty of Medicine, Friedrich Alexander University, Erlangen-Nuremberg, Erlangen, Germany.

4 Department of Anesthesiology and Critical Care, Klagenfurt Hospital, 9020 Klagenfurt, Austria

*Corresponding author

Neef V, Himmele C, Piekarski F, Blum LV, Hof L, Derwich W, Holubec T, Meybohm P, Choorapoikayil S. Effect of using smaller blood volume tubes and closed blood collection devices on total blood loss in patients undergoing major cardiac and vascular surgery. Can J Anaesth. 2024;71:213-23.
Prof Patrick Meybohm, MD: Patient Blood Management. https://www.patientbloodmanagement.de/pbm-informationen-fuer-aerzte/. Assessed 29 February 2024.
Smeenk FW, Janssen JD, Arends BJ, Harff GA, van den Bosch JA, Schönberger JP, Postmus PE. Effects of four different methods of sampling arterial blood and storage time on gas tensions and shunt calculation in the 100% oxygen test. Eur Respir J. 1997;10:910-3.
Prottengeier J, Jess N, Harig F, Gall C, Schmidt J, Birkholz T. Can we rely on out-of-hospital blood samples? A prospective interventional study on the pre-analytical stability of blood samples under prehospital emergency medicine conditions. Scand J Trauma Resusc Emerg Med. 2017;25:24.
Schinko H, Funk GC, Meschkat, M, Lamprecht B. Arterial blood gas analysis. Wien Klin. Wochenschr Educ. 2017;12: 115-30.
Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1:307-10.
German Medical Association. Guideline of the German Medical Association for the quality assurance of laboratory medical examinations - Rili-BÄK. Dtsch Arztebl 2014;111:A1583-1618.
Gruber MA, Felbermeir S, Lindner R, Kieninger M. Preanalytics: The (in-)stability of volatile POCT parameters and the homogeneity of blood in syringes at the market. Clin Chim Acta. 2016;457:18-23.
Knowles TP, Mullin RA, Hunter JA, Douce FH. Effects of syringe material, sample storage time, and temperature on blood gases and oxygen saturation in arterialized human blood samples. Respir Care. 2006;51:732-6.
Tyuma I. The Bohr effect and the Haldane effect in human hemoglobin. Jpn J Physiol. 1984;34:205-16.

No competing interests reported.

Pre-analytical validity of arterial blood gas samples: A prospective experimental study on changes derived from time delay and mechanical stress

Status:

Version 1

Abstract

Figures

Background

Material and methods

Aim, design, and setting

Blood samples

Test scenario

Data analysis and statistics

Results

Discussion

Conclusion

Abbreviations

Declarations

References

Additional Declarations

Status:

Version 1