Accurate haemoglobin (Hb) measurement is essential in managing critically ill patients, where it supports the diagnosis of anaemia, guides transfusion decisions, and assesses the overall oxygen-carrying capacity of blood. Traditionally, venous blood analyzed by central laboratory haematology analysers has served as the gold standard due to its superior precision and reliability. However, in high-acuity settings such as Intensive Care Units (ICUs), rapid decision-making necessitates the use of point-of-care (POC) tools, with arterial blood gas (ABG) analysers offering immediate results.

ABG analysers are now widely used in ICU and emergency departments to provide real-time analysis of parameters including blood gases, electrolytes, glucose, and haemoglobin. Critical care guidelines recommend round-the-clock ABG availability to meet the fast-paced demands of acute care environments [1]. Their ability to generate results quickly enhances workflow efficiency and supports prompt clinical interventions.

Despite these benefits, the accuracy of haemoglobin values obtained from ABG analysers compared to conventional lab methods remains controversial. Research has revealed discrepancies between these two modalities. Allardet-Servent et al. conducted a prospective observational study and found that ABG-derived Hb values differ significantly from central lab results, particularly in critical care patients, due to systematic bias and device-specific limitations [2]. Likewise, Nikhila et al. reported that while ABG analysers provide a rapid estimate of Hb levels, their accuracy may vary in specific clinical scenarios and should be confirmed by laboratory measurements [3]. A multicentre study by Triplett et al. further validated these findings, emphasizing the potential risks of relying solely on ABG analysers without confirmatory testing [4].

The physiological and methodological differences between arterial and venous samples may explain these discrepancies. Daae et al. found haematological variations, including Hb and hematocrit levels, between capillary and venous blood even in healthy adults [5]. This variation is further compounded in capillary samples, which may include interstitial and intracellular fluid contamination, leading to reduced measurement precision [6]. Conway et al. demonstrated that single-drop blood samples collected by skin puncture were associated with reduced precision in Hb measurements [7].

Moreover, frequent blood draws in ICU settings are not without consequence. Von Ahsen et al. highlighted the contribution of diagnostic and nondiagnostic phlebotomy to iatrogenic anaemia in ICU patients, especially when blood samples are repeatedly drawn in large volumes [8]. Hence, while ABG analysers offer the advantage of reducing turnaround time and sample volume, their use must be critically assessed in terms of diagnostic reliability.

Given these concerns, this study aims to assess the correlation and agreement between Hb values measured using ABG analysers and those obtained from conventional laboratory methods in adult ICU patients. By identifying the extent and sources of any discrepancies, this research seeks to support evidence-based use of ABG-derived haemoglobin in critical care practice.

Study Design

This research is a retrospective observational cross-sectional study conducted to assess the correlation between haemoglobin (Hb) values measured using an arterial blood gas (ABG) analyser and those obtained through conventional laboratory methods using venous samples.

Setting and Duration

The study was conducted in the Surgical Intensive Care Unit (ICU) at AIIMS Nagpur, over a one-month period. Data were collected from eligible patients who underwent both arterial and venous blood sampling as part of routine clinical management.

Study Population

Inclusion Criteria

• Adults aged 18–65 years.
• Both male and female patients.
• Patients for whom both arterial and venous blood samples were drawn as part of clinical care.

Exclusion Criteria

• Patients with diagnosed hemoglobinopathies.
• Haemolyzed samples or those compromised during collection.
• Patients with multi-organ dysfunction syndrome (MODS).
• Pregnant or lactating women.

Variables

The primary outcomes included haemoglobin values measured using arterial blood samples analyzed via an ABG analyser and venous samples analyzed through a standard central laboratory method.

Data Collection and Measurement Procedures

Arterial blood samples were analyzed immediately at bedside using an ABG analyser. Simultaneously, venous blood samples were collected and transported under standard conditions to the central laboratory, where haemoglobin estimation was carried out using an automated haematology analyser. All analysers were subject to routine quality control and calibration in accordance with institutional protocols. Trained ICU personnel followed standardized procedures for both sample collection and analysis to minimize variability.

Bias Control

To reduce selection bias, all eligible patients during the study period were recruited consecutively. Measurement bias was minimized through strict adherence to standard operating procedures and regular quality control checks of both analysers. Personnel involved in the process were trained uniformly to ensure consistency in handling and processing of samples.

Sample Size Calculation

Based on prior studies that reported mean haemoglobin values of 12.47 ± 1.41 g/dL for ABG analysis and 10.72 ± 1.17 g/dL for conventional laboratory methods, and assuming 90% power and a 95% confidence level, a minimum sample size of 8 was calculated to detect a statistically significant difference. To improve the reliability and generalizability of the findings, a total of 40 patients were included in the study.

Statistical Analysis

Data were first compiled in Microsoft Excel and then analyzed using SPSS version 20.0. The normality of distribution for haemoglobin values was assessed. To compare the mean haemoglobin levels obtained via ABG and laboratory methods, a paired t-test was performed. The Pearson correlation coefficient was calculated to assess the strength and direction of the relationship between the two measurement methods. Additionally, a simple linear regression model was employed to develop a predictive equation, allowing estimation of laboratory haemoglobin values from ABG readings. A p-value of less than 0.05 was considered statistically significant for all tests conducted.

A total of 40 paired haemoglobin (Hb) measurements were included in the final analysis after the exclusion of incomplete records. Each patient had haemoglobin levels measured both by an arterial blood gas (ABG) analyser and a conventional venous laboratory method.

The primary objective was to compare the haemoglobin values obtained from the two different methods, assess the strength of correlation, develop a predictive model, and evaluate the agreement between measurements.

Study Population

A total of 40 arterial and venous paired haemoglobin samples were analyzed. All included samples were eligible based on completeness of paired measurements and absence of data anomalies.

Comparison of Haemoglobin Measurements

The mean arterial haemoglobin (ABG) was 9.54 ± 2.16 g/dL, while the mean venous haemoglobin (laboratory method) was 10.00 ± 2.57 g/dL. The mean difference between ABG and lab values was -0.46 ± 2.09 g/dL (95% CI: -1.16 to 0.23). A paired t-test demonstrated that the difference in means was not statistically significant (p = 0.187) (Table 1).

Correlation between Arterial and Venous Hemoglobin Values

A moderate positive correlation was observed between hemoglobin values measured by the arterial blood gas analyzer and those obtained through the conventional laboratory method.
The Pearson correlation coefficient was r = 0.623 with a p-value < 0.001, indicating a statistically significant association between the two measurement methods (Figure 1).

Prediction Model for Laboratory Haemoglobin Values
A simple linear regression analysis was performed to predict venous haemoglobin levels based on arterial haemoglobin values.
The regression equation derived was:
Venous Haemoglobin (g/dL) = 2.94 + 0.74 × Arterial Haemoglobin (g/dL).

The model demonstrated a moderate strength of association with an R-squared value of 0.388, indicating that approximately 38.8% of the variance in venous haemoglobin could be explained by arterial haemoglobin readings.

The arterial haemoglobin value was a statistically significant predictor (p < 0.001), whereas the intercept term did not reach statistical significance (p = 0.064) (Table 2).

Agreement Analysis between Arterial and Venous Hemoglobin Values

Agreement between the arterial and venous hemoglobin measurements was evaluated using a Bland-Altman analysis.

The mean bias (arterial minus venous hemoglobin) was -0.46 g/dL, with a standard deviation of 2.09 g/dL.
The upper and lower limits of agreement were calculated to be +3.63 g/dL and -4.55 g/dL, respectively.

These findings suggest that although the mean difference between methods is small, there is considerable variability in individual patient measurements. This degree of variability may limit the interchangeability of ABG and laboratory hemoglobin measurements in critical clinical decisions (Figure 2).

Summary of Results and Inferences

1. No significant mean difference was observed between ABG and laboratory haemoglobin values, implying that on a population level, ABG measurements may approximate venous laboratory measurements.
2. Moderate positive correlation between ABG and venous haemoglobin suggests that the two methods are related, but not perfectly interchangeable for individual patient assessment.
3. Predictive ability is moderate: Arterial haemoglobin can estimate venous values to some extent, but a large portion of variability remains unexplained.
4. Clinical caution is warranted:
   The wide limits of agreement observed in the Bland-Altman analysis imply that for critical clinical decisions — particularly those requiring precise haemoglobin thresholds (e.g., transfusion triggers) — reliance solely on ABG haemoglobin values may not be advisable without confirmatory venous testing.

This retrospective observational study evaluated the correlation and agreement between arterial haemoglobin (Hb) values measured via arterial blood gas (ABG) analysers and venous Hb values from a conventional laboratory method. While no statistically significant difference was found between the mean values of the two methods, only a moderate positive correlation was observed. Regression analysis indicated modest predictive capability, and Bland-Altman analysis demonstrated wide limits of agreement, suggesting limited interchangeability in individual patient cases.

Our results align partially with prior literature, highlighting both the strengths and limitations of ABG-derived haemoglobin values in critically ill patients.

Rajan et al. (2022) [9] reported consistently higher ABG haemoglobin values compared to venous lab values across all perioperative stages of head-and-neck surgeries, with a mean difference of approximately 1 g/dL. This contrasts with our study, where the mean difference was -0.46 g/dL, though not statistically significant. This variation may reflect procedural or population-specific factors.

Ray et al. (2002) [10] and R. et al. (2016) [11] found a very high correlation (r² = 0.98) between ABG and lab values, with small mean differences (-4.3 g/L), suggesting high concordance in some settings. However, these studies also acknowledged that a small proportion of values fell outside clinically acceptable ranges — a concern echoed in our Bland-Altman analysis, which showed a bias of -0.46 g/dL but wide limits of agreement (−4.55 to +3.63 g/dL), underscoring potential risk in borderline clinical decisions.

Al Enezi et al. (2015)[12] also reported a strong positive correlation (p < 0.01) between ABG and venous Hb levels and proposed ABG analysis as a valid alternative in most cases. However, they cautioned against its use in rare scenarios requiring exact precision, a nuance supported by our findings.

Gibbons et al. (2019) [13] in their study involving emergency department patients, found haemoglobin biases up to −1.6 g/dL with ABG vs. lab analysis, and confirmed >95% of values within acceptable limits. While they concluded ABG to be clinically acceptable for time-sensitive decisions, they acknowledged consistent underestimation, which matches our observed mean bias.

Zhang et al. (2015) [14] reported no statistically significant difference in haemoglobin between ABG and lab methods, suggesting their interchangeability within accepted bias limits. However, our study adds nuance by showing that while the average difference may be negligible, individual variation remains substantial — potentially problematic in critically ill or transfusion-threshold cases.

Together, these findings reinforce that while ABG haemoglobin estimations may serve as rapid proxies, clinical context, patient acuity, and decision-critical thresholds must guide their interpretation.

Clinical Implications

ABG-derived haemoglobin values may be effectively used for rapid screening, trend monitoring, and routine surveillance in high-acuity settings like ICUs. However, due to the observed variability and wide limits of agreement, venous laboratory confirmation is recommended when precise quantification is necessary — such as in anaemia diagnosis or transfusion decision-making.

Strength and Limitations

This study benefits from a well-controlled paired design, limiting intra-patient variation and enhancing reliability. Calibrated equipment and consistent methodology further strengthen internal validity.

Limitations include the retrospective design and absence of clinical covariates (e.g., comorbidities, hydration status), which could influence haemoglobin variability. Additionally, our single-centre setting may limit generalizability. A relatively small sample size may also reduce power for subgroup analysis.

Future Directions

Future research should explore ABG-lab agreement across specific clinical subgroups (e.g., haemorrhage, sepsis, transfusion), integrate timing and handling data, and test correction models or device-specific adjustments to enhance ABG accuracy. Multi-centre, prospective studies would also improve external validity.

Although ABG analysers provide haemoglobin estimates close to laboratory values on average, moderate correlation and wide individual variation limit their standalone use. ABG-derived haemoglobin should be interpreted cautiously and confirmed with venous testing in decisions requiring precise haemoglobin thresholds.

1. Society of Critical Care Medicine. (1991). Guidelines for standards of care for patients with acute respiratory failure on mechanical ventilatory support. Critical Care Medicine, 19(2), 275–278. https://pubmed.ncbi.nlm.nih.gov/1994977/
2. Allardet-Servent, J., Lebsir, M., Dubroca, C., Fabrigoule, M., Jordana, S., Signouret, T., et al. (2017). Point-of-care versus central laboratory measurements of hemoglobin, hematocrit, glucose, bicarbonate and electrolytes: A prospective observational study in critically ill patients. PLOS ONE, 12(1), e0169593. https://doi.org/10.1371/journal.pone.0169593
3. Nikhila, K., Kiranmai, D., & Sreedevi, B. (2016). Estimation of Haemoglobin with Arterial Blood Gas Analyzer Compared to Conventional Laboratory Methods in Intensive Care Unit. ResearchGate. https://www.researchgate.net/publication/348419089
4. Triplett, K. E., Wibrow, B. A., Norman, R., Hince, D. A., Hardy, L. E., Tan, S., Ho, K. M., & Anstey, M. H. (2019). Can the blood gas analyser results be believed? A prospective multicentre study comparing haemoglobin, sodium and potassium measurements by blood gas analysers and laboratory auto-analysers. British Journal of Biomedical Science, 70(3), 103–109. https://doi.org/10.1080/09674845.2013.11740810
5. Daae, L. N., Halvorsen, S., Mathisen, P. M., & Mironska, K. (1988). A comparison between haematological parameters in 'capillary' and venous blood from healthy adults. Scandinavian Journal of Clinical and Laboratory Investigation, 48(7), 723–726. https://doi.org/10.3109/00365518809085783
6. Tatsumi, N., Miwa, S., & Lewis, S. M. (2002). Specimen collection, storage, and transmission to the laboratory for hematological tests. International Journal of Hematology, 75(3), 261–268. https://doi.org/10.1007/BF02982073
7. Conway, A. M., Hinchliffe, R. F., Earland, J., & Anderson, L. M. (1998). Measurement of haemoglobin using single drops of skin puncture blood: is precision acceptable? Journal of Clinical Pathology, 51(3), 248–250. https://doi.org/10.1136/jcp.51.3.248
8. Von Ahsen, N., Müller, C., Serke, S., Frei, U., & Eckardt, K. U. (1999). Important role of nondiagnostic blood loss and blunted erythropoietic response in the anemia of medical intensive care patients. Critical Care Medicine, 27(12), 2630–2639. https://doi.org/10.1097/00003246-199912000-00020
9. Rajan S, Tosh P, Isaac M, Sasikumar NK, Subramanian A, Paul J, Kumar L. Comparison of Hemoglobin Values Obtained by Arterial Blood Gas Analysis versus Laboratory Method during Major Head-and-Neck Surgeries. Anesth Essays Res. 2022 Jan-Mar;16(1):84–88. doi:10.4103/aer.aer_67_22. PMID: 36249132; PMCID: PMC9558660.
10. Ray JG, Post JR, Hamielec C. Use of a rapid arterial blood gas analyzer to estimate blood hemoglobin concentration among critically ill adults. Crit Care. 2002 Feb;6(1):72–75. doi:10.1186/cc1456. PMID: 11940269; PMCID: PMC83849.
11. R N, Madhusudhana R, Sujatha M, N S, K D. Estimation of haemoglobin with arterial blood gas analyzer compared to conventional laboratory methods in intensive care unit. IOSR J Dent Med Sci. 2016;15(8):5–7. doi:10.9790/0853-1508090507.
12. Al Enezi F, Al Anazi A, Al Mutairi M, Al Shahrani M, Qureshi S, Karthika M. Arterial and venous estimation of hemoglobin: A correlational study. J Hematol. 2015;4(3):187–192. doi:10.14740/jh224e.
13. Gibbons M, Klim S, Mantzaris A, Dillon O, Kelly AM. How closely do blood gas electrolytes and haemoglobin agree with serum values in adult emergency department patients: An observational study. Emerg Med Australas. 2019 Apr;31(2):241–246.
14. Zhang JB, Lin J, Zhao XD. Analysis of bias in measurements of potassium, sodium and hemoglobin by an emergency department-based blood gas analyzer relative to hospital laboratory autoanalyzer results. PLoS One. 2015 Apr 7;10(4):e0122383. doi:10.1371/journal.pone.0122383. PMID: 25849375; PMCID: PMC4388527