Sage Journals: Discover world-class research

Abstract

Background

Studies comparing venous total carbon dioxide (tCO₂) and standard hydrogen carbonate (HCO₃^–(P,st)) has shown diverse results, and it is debatable whether these two parameters can be used interchangeably for workup of acid–base disorders in a hospital setting.

Method

All patients with an HCO₃^–(P,st) requisition from any department at Odense University Hospital between 11th May 2021 and 1st June 2021 had tCO₂ and HCO₃^–(P,st) analysed simultaneously. TCO₂ was measured on Cobas® 8000, c702 module, while HCO₃^–(P,st) was calculated based on measurements on ABL835 Flex.

Results

From 1210 patients, mean (standard deviation (SD)) was 22.9 (3.7) mmol/L for tCO₂ and 22.5 (2.9) mmol/L for HCO₃^–(P,st). TCO₂ range was 10.1–42.3 mmol/L and 11.7–41.4 mmol/L for HCO₃^–(P,st). Linear regression showed that tCO₂ (mmol/L) = −2.90 + 1.15 × HCO₃^–(P,st) (mmol/L) with R² = 0.81. Bias (mean (SD) difference) between tCO₂ and HCO₃^–(P,st)) was 0.4 (1.7) mmol/L with a −5.0–9.6 mmol/L range. Limits of agreement was −2.90–3.70 mmol/L. Comparison of classification within, above or below reference interval for tCO₂ and HCO₃^–(P,st) showed that 984 samples (81%) retained their classification. Only one sample (0.1%) would be severely misclassified (outside the respective reference intervals) if HCO₃^–(P,st) was considered the gold standard. Of the samples investigated, 46.1% had a mean difference between tCO₂ and HCO₃^–(P,st) of 0–1 mmol/L and 30.3% had 1.1–2.0 mmol/L.

Conclusions

Our results indicate that venous tCO₂ and venous HCO₃^–(P,st) can be used interchangeably in a hospital setting for workup of acid–base disorders.

Keywords

Carbon dioxide hydrogen carbonates plasma hospital laboratory acid-base disorders

Introduction

The carbon dioxide (CO₂)–hydrogen carbonate buffer system, described by the Henderson–Hasselbalch equation, is crucial for pH control in the human body.^1,2 Standard hydrogen carbonate (HCO₃^–(P,st)) constitute around 95% of the total CO₂ (tCO₂) volume in human blood.^3,4 As diseases of the kidneys, liver or lungs can cause changes in blood pH,^1,2 measurement of tCO₂ or HCO₃^–(P,st) are important to detect and monitor acid-base disorders. In both chronic and acute patients, analysis of tCO₂ or HCO₃^–(P,st) provides important clinical information as patients can present mild or few symptoms. Furthermore, tCO₂ or HCO₃^–(P,st) can be used to monitor treatment response and disease regression or progression.

The tCO₂ analysis measures all CO₂ forms in the blood (hydrogen carbonate, carbonate, dissolved CO₂ etc.),⁵ while the HCO₃^–(P,st) analysis is the calculated amount of hydrogen carbonate in the blood under standard conditions.⁶ Analysis of tCO₂ can be performed on automatic high-throughput equipment connected to an automated track, while HCO₃^–(P,st) often is performed on stand-alone blood gas analysers requiring manual handling. Therefore, the time from phlebotomy to analysis is potentially longer for tCO₂ and with longer uncapping time but is cheaper and less prone to human errors. Unfortunately, studies comparing tCO₂ and HCO₃^–(P,st) has shown conflicting results,^3,7,8 and whether these two analyses can be used interchangeably is therefore questionable. The purpose of this study was to compare venous tCO₂ and venous HCO₃^–(P,st) from a broad spectrum of patients to determine, whether venous tCO₂ can substitute venous HCO₃^–(P,st) in a hospital setting for workup of acid–base disorders.

Methods

We performed a retrospective comparison study of HCO₃^–(P,st) and tCO₂. All patients with a HCO₃^–(P,st) requisition from any department at Odense University Hospital in the period 11th May 2021 to 1st June 2021 were included. For all included patients, tCO₂ was analysed simultaneously with standard HCO₃^–(P,st) as described below.

Phlebotomy was performed by trained phlebotomists according to the ISO15189 accredited standard operating procedure, and samples were received at the laboratory within 30 min.

After phlebotomy, a Li-heparin vacutainer® (Becton Dickinson, Franklin Lakes, NJ, USA) for tCO₂ was centrifuged and uncapped at the automated laboratory track (GLP, Abbott, Chicago, IL, USA), and tCO₂ was measured immediately on a Cobas® 8000, c702 module (Roche, Basel, Switzerland), connected to the track with dedicated reagents. Based on previous findings of tCO₂ stability⁹ and a small pilot project (n = 40) performed at our department, the timespan allowed for Li-heparin plasma samples to be uncapped until tCO₂ measurement was limited to 60 min.

Simultaneously with the Li-heparin vacutainer, a Na-Fluoride-Citrate vacutainer® (BD, Franklin Lakes, NJ, USA) was collected for HCO₃^–(P,st) analysis and handled manually at the laboratory. After 3 min of mixing, the sample was uncapped and immediately analysed on ABL835 Flex (Radiometer Medical, Copenhagen, Denmark) followed by calculation of HCO₃^–(P,st) using the embedded equation: HCO₃^–(P,st) = 24.47 + 0.919 × Z + Z x a´ x (Z-8), where Z = cBase(B)-0.3062 × ctHb x (1-sO₂) and a´ = 4.04 × 10⁻³ + 4.25 × 10⁻⁴ x ctHb.¹⁰

Limit of detection was 2 mmol/L for tCO₂. Coefficient of variation (CV%) at our laboratory was 2.3% at 17 mmol/L tCO₂ and 2.1% at 39 mmol/L tCO₂. Reference interval (RI) for tCO₂ was 22–29 mmol/L¹¹ and 22–27 mmol/L for HCO₃^–(P,st).¹² Our a priori demand for a clinically acceptable limit of agreement (LoA) in the Bland–Altman plot was 7.3%.^13,14

According to local ethical guidelines, the data used in this study constitutes a quality project and does therefore not require ethical approval from the Ethical Committee of Region Southern Denmark. All data obtained were completely anonymised, and according to the Danish National Committee on Health Research Ethics permissions from the Danish authorities were not required.¹⁵ Therefore, only data regarding tCO₂ and HCO₃^–(P,st), including the department requesting the analysis, was available. Samples included in this study were requested for routine HCO₃^–(P,st) analysis and would otherwise be discarded afterwards. The tCO₂ analysis did not require an additional sample or phlebotomy. Furthermore, the tCO₂ result was not visible to the requesting physician and therefore did not influence any clinical decisions.

Results

In total, tCO₂ and HCO₃^–(P,st) were analysed simultaneously in samples from 1210 patients. Of note, 702 (58%) of the samples were retrieved from the Nephrology Department, while the remaining samples were from a wide variety of clinical departments.

The Anderson-Darling test on tCO₂ and HCO₃^–(P,st) showed that data was not normally distributed (p-value: <0.05). However, the residuals on the linear regression were normally distributed and we found homoscedasticity. Mean (standard deviation (SD)) was 22.9 (3.7) mmol/L for tCO₂ and 22.5 (2.9) mmol/L for HCO₃^–(P,st). Range was 10.1–42.3 mmol/L for tCO₂ and 11.7–41.4 mmol/L for HCO₃^–(P,st). Bias (mean (SD) difference) between tCO₂ and HCO₃^–(P,st)) was 0.4 (1.7) mmol/L with a −5.0-9.6 mmol/L range. Of the 1210 samples, 46.1% had a mean difference between tCO₂ and HCO₃^–(P,st) of 0–1 mmol/L, 30.3% had 1.1–2.0 mmol/L, 16.5% had 2.1–3.0 mmol/L, 5.3% had 3.1–4.0 mmol/L, 1.4% had 4.1–5.0 mmol/L, and 0.3% had >5.0 mmol/L. Bland-Altman difference plot showed a concentration-dependent distribution, where samples in the low concentration range had higher HCO₃^–(P,st) concentrations than tCO₂, while samples in the high concentration range had higher tCO₂ concentrations than HCO₃^–(P,st) (Figure 1).

Figure 1.

Bland–Altman difference plot showing the comparison of total CO₂ (tCO₂) and standard hydrogen carbonate (HCO₃^–(P,st)) measured simultaneously in patients from any department at Odense University Hospital between 11^th May and 1^st June 2021. N = 1210. Mean value (0.40 mmol/L) is shown as a black line and 95% limit of agreement (LoA) as dotted lines (−2.90–3.70 mmol/L).

Comparison of classification within, above or below established RI for tCO₂ and HCO₃^–(P,st), respectively, revealed that 984 samples (81%) retained their classification. If HCO₃^–(P,st) was considered the gold standard, one sample (0.1%) would be severely misclassified, defined as below RI for HCO3–(P,st) and above RI for tCO₂, when using the tCO₂ analysis result (Table 1), while no samples would be severely misclassified if tCO₂ was considered the gold standard. Linear regression showed that tCO₂ (mmol/L) = −2.90 + 1.15 × HCO₃^–(P,st) (mmol/L) with R² = 0.81, and 95% CI for intercept −3.61 to −2.19 and 1.12–1.18 for slope (Figure 2).

Table 1.

Classification within, above or below reference interval (RI) for 1210 samples of total CO2 (tCO₂) and standard hydrogen carbonate (HCO3–(P,st)) measured simultaneously in patients from any department at Odense University Hospital between 11th May and 1st June 2021. RI for tCO₂ was 22–29 mmol/L¹¹ and 22–27 mmol/L for HCO₃^–(P,st).¹²

	Standard hydrogen carbonate (HCO₃^–(P,st))
Total CO₂ (tCO₂)	RI	<22 mmol/L	22–27 mmol/L	>27 mmol/L
	< 22 mmol/L	395 (33%)	87 (7%)	0 (0%)
	22-29 mmol/L	99 (8%)	543 (45%)	25 (2%)
	> 29 mmol/L	1 (0.1%)	14 (1%)	46 (4%)

Figure 2.

Linear regression showing the correlation between total CO₂ (tCO₂) and standard hydrogen carbonate (HCO₃^–(P,st)) measured simultaneously in patients from any department at Odense University Hospital between 11th May and 1st June 2021. N = 1210. Regression formula: tCO₂ (mmol/L) = −2.90 + 1.15 × HCO₃^–(P,st) (mmol/L) with R² = 0.81, and 95% CI for intercept −3.61 to −2.19 and 1.12–1.18 for slope.

Discussion

Since the tCO₂ analysis can be performed automatically and therefore is less laborious and cheaper than HCO₃^–(P,st), our interest was to determine whether tCO₂ can safely substitute HCO₃^–(P,st) in a hospital setting. In 1210 patient samples, we found a linear relationship between tCO₂ and HCO₃^–(P,st) compatible with previous findings.^3,7 The R²-values however differ between the studies, which could be due to differences in equipment, analysis method, and cohort; e.g. the study by Hirai et al.⁷ was based on pre-dialysis patients with chronic kidney disease. Despite a strong linear relationship, Nasir et al.³ concluded that tCO₂ could not substitute HCO₃^–(P,st) based on Story and Poustie¹⁶’s criterion . This criterion states that plasma HCO₃^–(P,st) concentrations is clinically important if bias is greater than ±1 mmol/L and LoA is wider than bias ±2 mmol/L (span of maximum 4 mmol/L) . Similar to Nasir et al.,³ our findings did not fulfil the criterion: Although our data only indicated a bias of 0.40 mmol/L, LoA spanned 6.60 mmol/L (−2.90–3.70). We would however like to question the usability of Story and Poustie’s criterion in this context: It originates from a comparison study of two different arterial HCO₃^–(P,st) analyses on 102 intensive care patients, and it is therefore safe to presume that it is not directly applicable to venous samples from patients attending a hospital in general.

Majority of the samples had a mean difference between tCO₂ and HCO₃^–(P,st) within ±3.0 mmol/L and 46% of the samples within ±1.0 mmol/L. We found that samples in the low concentration range had higher HCO₃^–(P,st) concentrations than tCO₂, while samples in the high concentration range had higher tCO₂ concentrations compared to HCO₃^–(P,st). This could indicate that tCO₂ and HCO₃^–(P,st) results are not directly equivalent, and an analytical or pre-analytical bias could be present, and the Bland–Altman plot reveals that there are no systematic bias. If looking solely at the mean difference distribution it could be argued that there is an analytical uncertainty. However, when looking at the mean difference from a clinical point of view, the agreement was deemed sufficient as changes of 1–3 mmol/L within the RI was considered clinically irrelevant. For results outside the RI, we recommend to perform additional analysis (e.g. pH), especially when results are below RI, where we found that tCO₂ tended to be lower than HCO₃^–(P,st), and therefore potentially could help early detection of metabolic acidosis or a compensated respiratory alkalosis. Furthermore, the concentration-dependent difference could indicate that further studies on the RI for tCO₂ and HCO₃^–(P,st) is needed. In a clinical context, this indicates that for acute patients with symptoms of respiratory or metabolic acidosis or alkalosis, it could be beneficial to measure pH simultaneous with tCO₂, as results in acute patients tends to be outside RI. However, in chronic patients with a known disease who needs monitoring, venous tCO₂ could be part of a standard surveillance, hopefully decreasing the number of phlebotomies performed.

We measured HCO₃^–(P,st) immediately after uncapping the samples, whereas tCO₂ was measured up to 60 min after uncapping. Uncapping causes diffusion of gases from the sample and therefore reduces the tCO₂ level. This factor could potentially increase bias and cause misclassification of results within, above or below RI. However, Kirschbaum⁹ found a 5% change in tCO₂ after 60 min exposure to room air, and our small pilot study of 40 samples found a 7% change, which we deemed was within clinically acceptable range.

Of note, it is important to emphasize that comparison of biomarkers requires a gold standard to assure that a misclassification is not caused by superiority of the new biomarker. In our case, the clinical presentation along with pH and pCO2 could be considered the gold standard, as measurement of HCO₃^–(P,st) also has its limitations and can be affected by recent treatment (eg dialysis). Unfortunately, data on clinical presentation were not available. However, tCO₂ and HCO₃^–(P,st) is widely used in patients with risk of, or clinical symptoms indicating, acidosis or alkalosis, where pH is also a key parameter. A large retrospective study of 21,586 samples showed a very strong correlation (R = 0.99) between real pH and pH estimated from tCO₂.¹⁷ When comparing tCO₂ and pH for classification of acidosis and alkalosis in patients on continuous kidney replacement therapy, tCO₂ is reported to have a high misclassification risk,⁶ but whether this also holds true for HCO₃^–(P,st) is unknown and warrants further investigation.

Arterial and venous HCO₃^–(P,st) has been found reasonably compatible in clinical settings,^18,19 and although the agreement for pCO₂ was poor, direct comparison of venous and arterial tCO₂ is sparsely evaluated. A study of 151 paired arterial and venous samples found an acceptable correlation between venous tCO₂ and arterial HCO₃^–(P,st).²⁰ Our study is based on venous tCO₂ and venous HCO₃^–(P,st), but in emergency situations, we recommend use of arterial blood gas analysis, which also provides a variety of other critical parameters.

In conclusion, our results indicate that venous tCO₂ and venous HCO₃^–(P,st) can be used interchangeably in a hospital setting for workup of acid-base disorders. In a clinical context, it would be beneficial to measure pH simultaneous with tCO₂ in acute patients, while in chronic patients with a known disease who needs monitoring for acid-base disorders, venous tCO₂ could be part of a standard surveillance.

Footnotes

Acknowledgements

All the authors have accepted full responsibility for the entire content of this submitted manuscript and approved the submission. We would like to thank Louise Falk Nielsen, Johanne Rullhøj Knudsen, Kathrine Keil and Pernille Winther Thrane for their contribution with performing the pilot study on tCO2 stability.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Ethical approval

According to local ethical guidelines, the completely anonymised data used in this study constitutes a quality project and does not require ethical approval.

Guarantor

Mads Nybo.

Contributorship

RSH, JR, MBS, LHJ and MN: Protocol development, performing the study, obtaining data, analysis of data, editing and critical review of the manuscript RSH: Researched literature, Wrote paper draft, MBS and LHJ: Performed the figures.

ORCID iDs

Rasmus Søgaard Hansen

Mads Nybo

References

Hamm

Nakhoul

Hering-Smith

. Acid-base homeostasis. Clin J Am Soc Nephrol 2015; 10: 2232–2242.

Berend

de Vries

Gans

. Physiological approach to assessment of acid-base disturbances. N Engl J Med 2014; 371: 1434–1445.

Nasir

Sthaneshwar

Yunus

, et al. Comparing measured total carbon dioxide and calculated bicarbonate. Malays J Pathol 2010; 32: 21–26.

Kraut

Madias

. Re-evaluation of the normal range of serum total CO2 concentration. Clin J Am Soc Nephrol 2018; 13: 343–347.

Chung

Lee

Chun

, et al. Changes in total CO2 measurement according to reagent cassette rotation in chemistry autoanalyzers. Ann Clin Lab Sci 2009; 39: 150–154.

Lawrie

Golda

. Actual or standard bicarbonate. Lancet 1979; 2: 201–202.

Hirai

Minato

Kaneko

, et al. Approximation of bicarbonate concentration using serum total carbon dioxide concentration in patients with non-dialysis chronic kidney disease. Kidney Res Clin Pract 2019; 38: 326–335.

Woodell

Webster

Mehta

, et al. Total carbon dioxide versus pH for determining acid-base status in patients on continuous kidney replacement therapy: a cohort study. Am J Kidney Dis 2021; 77: 305–307.

Kirschbaum

Loss of carbon dioxide from serum samples exposed to air. Effect on blood gas parameters and strong ions. Clin Chim Acta 2003; 334: 241–244.

10.

Radiometer Medical ApS . ABL800 FLEX Reference Manual. 2012. 201206B.

11.

Tietz

. Clinical guide to laboratory tests. 3rd ed. Philadelphia (PA): W.B. Saunders; 1995.

12.

Olesen

. Kompendium i Laboratoriemedicin [Compendium of Laboratory Medicine]. Copenhagen: Amtsrådsforeningen i Danmark; 1988.

13.

Giavarina

. Understanding bland altman analysis. Biochem Med (Zagreb) 2015; 25: 141–151.

14.

Westgard

. WestgardQC. https://www.westgard.com/ (accessed 21 December 2022).

15.

Komitéloven . Bekendtgørelse af lov om videnskabsetisk behandling af sundhedsvidenskabelige forskningsprojekter [Act of the law on health research ethical treatment of health research projects], https://www.retsinformation.dk/eli/lta/2017/1083#idfae75201-60d1-46a6-bef7-ccd4ce8f39aa (2017, LBK nr. 1083 af 15/09/2017, Accessed 04 May 2022).

16.

Story

Poustie

. Agreement between two plasma bicarbonate assays in critically ill patients. Anaesth Intensive Care 2000; 28: 399–402.

17.

Bulucu

Çakar

Kurt

, et al. Blood total carbon dioxide content and bicarbonate can be used together to predict blood pH correctly in venous blood samples. Ren Fail 2014; 36: 145–146.

18.

Bloom

Grundlingh

Bestwick

, et al. The role of venous blood gas in the emergency department: a systematic review and meta-analysis. Eur J Emerg Med 2014; 21: 81–88.

19.

Ekström

Engblom

Ilic

, et al. Calculated arterial blood gas values from a venous sample and pulse oximetry: clinical validation. PLoS One 2019; 14: e0215413.

20.

Kim

Park

Shin

, et al. Correlation between peripheral venous and arterial blood gas measurements in patients admitted to the intensive care unit: a single-center study. Kidney Res Clin Pract 2013; 32: 32–38.

Comparison of plasma total carbon dioxide and standard hydrogen carbonate in a hospital laboratory setting

Abstract

Background

Method

Results

Conclusions

Keywords

Introduction

Methods

Results

Discussion

Footnotes

Acknowledgements

Declaration of conflicting interests

Funding

Ethical approval

Guarantor

Contributorship

ORCID iDs

References