Effects of Hyperventilation on Venous-Arterial Bicarbonate Concentration Difference : A Possible Pitfall in Venous Blood Gas Analysis

Objectives: Recent reports on venous blood gas analysis have shown that venous bicarbonate concentration is useful in the evaluation of the body acid-base status. Most of these reports have been based on the Bland-Altman analysis comparing arterial and venous blood gas values. We intended to elucidate any factors that decrease the agreement between venous and arterial bicarbonate concentrations, which might impair the usefulness of venous blood gas analysis. Methods: Healthy volunteers and patients with various diseases (n = 141) were evaluated by simultaneous arterial and venous blood sampling and Bland-Altman analysis. The venous-arterial bicarbonate concentration difference was compared between healthy volunteers and untreated respiratory alkalosis patients. Intentional hyperventilation (30 or 60 breaths/min, for 3 min) was also performed on 6 healthy volunteers and the venous-arterial bicarbonate concentration difference was evaluated. Results: The relative average bias in bicarbonate concentration was 2.00 mEq/l with venous bicarbonate higher than arterial bicarbonate with 95% limits of agreement of ±4.15 mEq/l. Hyperventilation challenges increased the venous-arterial bicarbonate concentration difference in an intensity-dependent manner. The venous-arterial bicarbonate concentration difference was higher in untreated respiratory alkalosis patients than in healthy volunteers (P < 0.01). Conclusion: Although venous bicarbonate may be useful to evaluate the body acid-base status, hyperventilation increases the venous-arterial bicarbonate concentration difference. Physicians should keep this phenomenon in mind.


Introduction
Since the pulse oximeter was invented and it became possible to evaluate the systemic oxygen level by measuring percutaneus oxygen saturation (SpO 2 ), peripheral venous blood gas analysis (VBGA) with simultaneous SpO 2 measurement has been considered useful as an alternative to arterial blood gas analysis (ABGA) [1][2][3].Indeed, physicians are now widely and even routinely performing VBGA with SpO 2 measurement instead of sampl-ing arterial blood, because VBGA is much easier and less invasive than ABGA especially in the youngest pediatric patients and in an emergency room.The agreement between variables on arterial and venous blood gas analysis has been well reviewed [1].The usual method to evaluate the agreement has been with the Bland-Altman analysis [4].
Previously we reported that intentional hyperventilation increased venous-arterial partial CO 2 pressure (PCO 2 ) differences and pH differences [2].We also reported that underestimation of respiratory alkalosis may occur with the "SpO 2 plus VBGA" method in untreated respiratory alkalosis patients [2].Here we evaluated the agreement in venous-arterial bicarbonate concentration ( )  measurements by the Bland-Altman analysis.The effects of intentional hyperventilation on the venousarterial 3 HCO − difference were also evaluated.In addition, the differences in healthy volunteers and in untreated respiratory alkalosis patients were also compared.

Subjects
The present study was approved by the Ethics Committees at Ohtawara Red Cross Hospital and the International University of Health and Welfare.141 subjects (95 males and 46 females, ranging from 16 to 91 years of age) were enrolled in this study after we obtaining their informed consent.Among these 141 subjects, 11 healthy volunteers and 130 patients with various diseases were included.Among these 130 patients, 13 patients with hyperventilation with PaCO 2 < 35 mmHg and arterial pH (pHa) > 7.45 without the treatment such as a paper bag re-breathing maneuver were included.

Blood Sampling and Gas Analysis
The brachial artery and the median vein were used for the blood sampling.Arterial and venous blood was sampled simultaneously with a small (1 ml) syringe containing heparin, and was immediately analyzed with an automatic blood gas analyzer (Rapidlab 840, Bayer Healthcare, Leverkusen, Germany, or Rapidlab 1265, Siemens Healthcare Diagnostics, Sudbury, United Kingdom).Blood sampling from healthy volunteers was done first at rest and then immediately after hyperventilation.The venous-arterial End-tidal PCO 2 (P ET CO 2 ) was measured with a gas analyzer (Respina IH26, NEC San-ei, Tokyo, Japan) [5].
Bicarbonate concentration was calculated by the following equation:

Protocols of Loading Maneuvers
In order to look at the effects of hyperventilation, the subjects breathed room air at a fixed rapid rate for 3 min.The breathing rate was changed from resting (11 -20 times per min) to 30 and then to 60 times per min.The timing of breathing was announced by a time keeper and the subjects followed his voice.The subjects were requested to keep the same tidal volume so that the P ET CO 2 was 30 ± 2 and 22 ± 2 mmHg during the 30 and 60 breaths/min hyperventilation maneuvers, respectively.

Statistical Analysis
Values are expressed as mean ± standard deviation unless indicated.We tested the linear correlation for the bicarbonate difference between ABGA and VBGA by the Spearman rank method, and compared the differences by Bland-Altman analysis [4].We used an analysis of variance with a Fisher post hoc multiple comparison for the evaluation of repeated measures between resting and intentional hyperventilation.An unpaired t-test (two-tail) was used for the comparison between healthy volunteers and patients.P < 0.05 was considered statistically significant.

Arterial and Venous Bicarbonate Concentration at Rest
The arterial and venous  and 95% limits of agreement of ±4.15 mEq/l.

Venous-Arterial Bicarbonate Concentration Difference in Patients with Hyperventilation
The ( )

Discussion
A meta-analysis and a review using Bland-Altman analysis reported that venous pH, bicarbonate and base excess have sufficient agreement to be clinically interchangeable for arterial values for patients in the emergency department [1,3].On the other hand, agreement between arterial and venous PCO 2 has been reported to be too poor and unpredictable in clinical usage as a one-off test, and PCO 2 might be useful to screen for arterial hypercarbia or monitor trends in PCO 2 for selected patients.
In untreated hyperventilation patients, we reported that the "VBGA plus SpO 2 " method may lead to underestimation or misdiagnosis of respiratory alkalosis [2].We speculated that the phenomenon could be attributed to the increase in the differences in venous-arterial PCO 2 and pH in the acute phase of hyperventilation.We also  speculated that the increases in these differences are due to the reduction of peripheral blood perfusion which was induced by hyperventilation-associated systemic vaso-constriction.Other reported factors that affect the differences in venous-arterial PCO 2 and pH are finger exercise and hypotension [2,6].
Here we evaluated ( ) and found that ( )  increases after the 3 min hyperventilation challenge in an intensity-dependent manner (Figure 2).( )  also increased in untreated respiratory alkalosis patients (Figure 3).The time course of ( )  after hyperventilation has not been well studied.Nevertheless, we suppose that ( )  is increased in the acute phase of hyperventilation (in the same time of developing respiratory alkalosis), but ( )  might decrease more swiftly than the nor- malization of respiratory alkalosis in arterial blood gas data.We experienced some already treated respiratory alkalosis patients using a paper-bag rebreathing maneuver without an increase in ( ) As for the Bland-Altman analysis data, our results for 3 HCO −     are similar to previous reports [7][8][9] (Table 1).The authors wrote that the agreement was acceptably narrow.Nevertheless, we feel that the 95% limits of agreement of ±5.05 mEq/l [9] are not so narrow.We Intensive care unit patients [10] cannot clearly say that our data of 95% limits of agreement of ±4.15 mEq/l are narrow or not narrow (important or not important).Our feeling is intermediate on this.Anyway, peripheral venous 3 HCO −     seems to be useful.As for the comparison between the arterial blood and "central" venous blood, Middleton et al. reported similar data [10] (Table 1).
Our report is the first that addresses how hyperventilation increases the difference between arterial and venous bicarbonate concentration.Physicians should keep this phenomenon in mind when performing venous blood gas analysis.

3 HCO
− data at rest (n = 141) are plotted in Figure 1.Data for both healthy volunteers and various patients are included.The relationship between the ABGA and VBGA was close for 3 HCO −     (r = 0.897, P = 3.85 × 10 −51 , Figure 1(a)).Bland-Altman plots are shown in Figure 1(b).The relative average bias of

3
healthy volunteers and from patients with untreated respiratory alkalosis (PaCO 2 < 35 mmHg and pH > 7.45) are shown in Figure

Figure 1 .
Figure 1.Blood gas data of healthy volunteers and patients with various diseases at rest (n = 141).(a) Relationship between arterial bicarbonate concentration ( ) −     3 a HCO and venous bicarbonate concentration ( ) −     3 v HCO is shown.

Figure 2 .
Figure 2. Intensity-response relationship between the level of hyperventilation and the venous-arterial bicarbonate concentration difference ( ) ( ) −     3 v a HCO − .The breathing rate was changed from resting (11 -20 times per min) to 30 and then to 60 times per min.Hyperventilation (Hyperv.)increased ( )