Abstract
The aim of the present study was to evaluate a method for calculating arterial values of pH, carbon dioxide tension (PCO2) and oxygen tension (PO2) from peripheral venous values.
In total, 40 patients were studied. Arterial and peripheral venous blood were sampled at a department of respiratory diseases. Arterial values were calculated from venous, and measured and calculated values of arterial pH, PCO2 and PO2 were compared.
Measured and calculated values of pH and PCO2 correlated well, with the difference between them having a very small bias and standard deviation (pH -0.001±0.013, PCO2 -0.09±0.28 kPa) within those considered acceptable for laboratory equipment and clinical practice. All but four patients had peripheral oxygen saturation (Sp,O2) ≤96%, and for these measured and calculated PO2 correlated well, with a difference such that the bias and standard deviation suggested that calculated &SetFont Typeface="43" Bold="1" Italic="1";P&SetFont Typeface="43" Bold="1";O2 may be clinically useful (&SetFont Typeface="43" Bold="1" Italic="1";P&SetFont Typeface="43" Bold="1";O2 0.11±0.53 kPa).
The present study evaluates a method for calculating arterial pH, carbon dioxide tension and oxygen tension from venous blood. It has been shown that arterial pH and carbon dioxide tension can be calculated precisely, and that oxygen tension can be calculated with reasonable precision in the vast majority of patients. This method might be useful in reducing the need for painful arterial punctures.
Arterial blood gas analysis is an important tool in assessing patient status. For patients residing in intensive care units (ICUs) with indwelling arterial catheters, the sampling of arterial blood is routine. In other departments blood is sampled via an arterial puncture with an increased risk of pain and haematoma 1–3.
Recently, studies have shown that values of acid–base status measured in peripheral venous blood correlate well with those measured in arterial blood, at least for values of pH, bicarbonate and carbon dioxide tension (PCO2) 4–6. Large numbers of peripheral venous samples are taken easily and safely in medical departments, therefore implying a role for peripheral venous blood in the monitoring of acid–base status. In addition, a method has been recently proposed that enables calculation of values of arterial acid–base and oxygenation status (i.e. pH, PCO2 and oxygen tension (PO2)) from measurements performed in the peripheral venous blood 7. This method has been shown to have good accuracy and precision when calculating arterial pH (bias±2sd: 0.002±0.027) or arterial PCO2 (bias±2sd: -0.04±0.52 kPa) 8. For calculation of PO2, precision was dependent on the oxygenation level, such that for peripheral oxygen saturation (Sp,O2) values >96% the method was not considered useful for calculating PO2, and for Sp,O2 ≤96% the accuracy and precision of calculated PO2 was 0.21±1.85 kPa (bias±2sd) 8.
The previous evaluation of this method was in patients admitted to the wards of either ICUs or a department of pulmonary medicine, and these patients were studied under experimental conditions 8. These experimental conditions differ from routine clinical practice in medical departments in several important respects. The majority of ICU patients have arterial catheters and are relatively immobile, making blood sampling simpler and less prone to error. Medical patients present as both admitted and outpatients, with admitted patients tending to have a poorer condition than outpatients, potentially giving larger arterial–venous differences in blood gas values and hence a greater signal-to-noise ratio for application of the conversion method. In experimental conditions, peripheral venous blood sampling was performed using a needle attached directly to a syringe, rather than the butterfly sampling needle and tube often used to sample venous blood in medical departments. In addition, the vast majority of patients included in the previous study, i.e. ICU patients, required supplementary oxygen as part of their clinical management 8, meaning that for the majority of patients Sp,O2 was >96%.
The purpose of the present study was to evaluate the method for conversion of peripheral venous values of acid–base and oxygenation status to arterial values in patients presenting at the department of pulmonary medicine, using sampling technology and blood samples routinely collected as part of clinical care.
MATERIALS AND METHODS
Patients
Arterial and peripheral venous blood was sampled from 40 patients (22 admitted and 18 outpatients) at the Dept of Respiratory Diseases at Aalborg Hospital (Aalborg, Denmark) of median age 68 yrs (range 35–93 yrs) and of whom 22 were females. Patients were all previously diagnosed with chronic lung disease resulting in hypercapnia and/or hypoxia, and were either acutely admitted or visiting the Dept of Respiratory Diseases for their biannual clinical assessment. The arterial acid–base and oxygenation status of the patients is summarised in table 1⇓.
Data acquisition
All patients had arterial blood sampled as a routine part of clinical care. These samples were drawn via an arterial puncture into a syringe suitable for blood gas analysis (PICO70; Radiometer AS, Brønshoj, Denmark) and processed immediately to obtain values of arterial acid–base, oxygen, metabolism and electrolyte status (ABL 800; Radiometer AS).
All patients had venous blood sampled as a routine part of clinical care. A butterfly needle (Multifly; Sarsted AG & Co., Nübrecht, Germany), was inserted into a vein at the elbow. This butterfly included a 20-cm long and 0.8-mm diameter tube connected to a sealed sample adaptor. Blood was sampled into a standard venous blood collection tube (S-Monovette; Sarsted AG & Co.) using the vacuum principle, and into a syringe suitable for blood gas analysis (PICO50; Radiometer AS), fitted with an adaptor (Membran adaptor; Sarsted AG & Co.) to enable sampling from the butterfly’s sample adaptor. Blood was analysed immediately for values of acid–base, oxygenation, metabolism and electrolyte status. Venous and arterial blood samples were taken within a 10-min period as part of normal clinical practice. In addition, the measurement of Sp,O2 was obtained from a pulse oximeter (Nonin Onyx 9500; Nonin Medical, Minneapolis, MN, USA) placed on an index finger. The data from two patients were excluded, due to sample error and the pulse oximeter running out of battery during blood sampling.
Blood sampling was performed in the present study as for routine clinical practice and, following contact with the local ethics committee, the investigators were informed that application for ethical approval was not required.
Venous to arterial conversion method
The following is a brief summarising description of the venous to arterial conversion method. Figure 1⇓ illustrates the method for calculating arterial acid–base and oxygenation status from measurements in the venous blood and pulse oximetry, described in detail previously 7. The method calculates arterial values using mathematical models 9 to simulate the transport of venous blood back through the tissues until simulated arterial oxygenation matches that measured by pulse oximetry. In this simulation, oxygen is added to and carbon dioxide removed from venous blood using a constant value of the respiratory quotient (RQ), selected in the present study to be 0.82 7, and no strong acid is removed from the blood on its simulated passage from the veins to the arteries, i.e. change in base excess from arterial to venous blood (ΔBEav) is 0 mmol·L−1.
Calculated arterial values of pH, PCO2 and PO2 were compared with the simultaneously measured arterial values using Bland–Altman plots and scatter plots of measured versus calculated. Values of bias and standard deviation between measured and calculated arterial values are reported, along with values of correlation coefficients (r2) and parameter values for regression lines.
RESULTS
Figure 2⇓ compares measured arterial and calculated arterial values using the venous to arterial conversion method. As illustrated in the Bland–Altman plots of figure 2⇓ and in table 2⇓, calculated values of arterial pH and PCO2 had very small bias and standard deviation, with these errors being within those considered acceptable for the performance of laboratory equipment (table 2⇓) and well within the limits of error acceptable in clinical practice 11, 12. Measured and calculated values of pH and PCO2 correlated extremely well (r2 for pH = 0.95, r2 for PCO2 = 0.98), with regression lines comparing well to the lines of identity. As illustrated in figure 2⇓, the error in calculated PO2 values was greater at higher levels of PO2. Table 2⇓ reports values of standard deviation for PO2 values corresponding to Sp,O2 ≤96% (34 patients) and Sp,O2 ≤98% (38 patients). These errors are outside those considered acceptable for the performance of laboratory equipment but may be useful in clinical practice. When data points with Sp,O2 ≥97% were excluded (fig. 2⇓), values of measured and calculated PO2 correlated well (r2 = 0.86).
Figure 3⇓ illustrates the sensitivity of calculations of arterial pH, PCO2 and PO2 to error in Sp,O2. Each of the plots represents simulation of one of these variables when a standard venous blood sample (pH 7.370, PCO2 6.1 kPa, PO2 5.5 kPa, oxygen saturation (SO2) 75%, haemoglobin concentration 9.3 mmol·L−1, methaemoglobin concentration 0 mmol·L−1, carboxyhaemoglobin concentration 0 mmol·L−1) is used to calculate arterial values using an Sp,O2 value ranging from 80 to 96%. Errors in Sp,O2 of ±4% gave rise to a small variation in calculated arterial pH with an sd of ±0.0025 (fig. 3a⇓), which was fairly constant over the range of Sp,O2 values and had a maximum of 0.005. A similar picture was seen in the error in calculation of arterial PCO2 (fig. 3b⇓). Errors in Sp,O2 of ±4% gave rise to a small variation in calculated arterial PCO2 with an sd of 0.06 kPa, which was again fairly constant over the range of Sp,O2 values and had a maximum sd of 0.12 kPa. The errors in calculation of arterial PO2 are illustrated in figure 3c⇓ and d, figure 3d⇓ being a magnified version of figure 3c⇓ (see Discussion section).
DISCUSSION
Arterial blood gas analysis is an important tool in assessing patient status, but outside the ICU blood sampling is typically performed from an arterial puncture and is associated with pain and increased risk of haematoma.
The present study provides further evaluation of a method for calculating arterial acid–base and oxygenation status from measurements taken in peripheral venous blood 7. Unlike a previous study 8, the present study has evaluated use of the method using routinely available blood samples in a patient group where arterial catheters are uncommon, i.e. patients presenting at a department of pulmonary medicine.
The current study has illustrated the potential for application of the method as part of the routine management of patients presenting at the department of pulmonary medicine. Calculation of arterial values could be performed with a similar precision to that for patients studied previously, despite the potential for increased error due to the logistics of blood sampling in routine clinical practice, and the use of butterfly-type needles and sample tubing. The accuracy and precision of calculated values of arterial pH and PCO2 were almost identical to those reported previously 8, and the precision was within laboratory acceptable performance criteria 12. For Sp,O2 values ≤96%, the method can calculate PO2 with an average sd of 0.53 kPa, which may be useful in clinical practice. The current sd is lower than that seen previously 8, where sd was 0.93 kPa. This is due to the lower values of arterial oxygenation seen in the patients presenting in the current study, with about 50% of patients having Sp,O2 ≤92%. For Sp,O2 values >96% it has previously been argued that the calculation of PO2 values becomes imprecise due to the flat shape of the oxygen dissociation curve at high levels of oxygenation. Only four of the current patients presented with Sp,O2 ≥97%, as illustrated in figure 2⇑. The small number of patients with high levels of oxygen indicates that the method might be applicable for calculation of arterial pH, PCO2 and PO2 in the vast majority of patients visiting pulmonary medical departments.
To understand whether the results presented in the current study would be reflected in a broader population, it is necessary to consider the sensitivity of calculations to measurement error and to the assumptions contained in the method. The method does not include calculation for the effects of varying blood temperature. All blood samples are measured at 37°C and all calculations performed at this temperature. Any necessary corrections for temperature can be performed using existing algorithms present in commercial blood gas analysers.
It has previously been shown that calculations performed using the method are insensitive to errors in measurement of venous blood gases, being linear over a range of input conditions and of an order of magnitude similar to the direct measurement of arterial blood 7. This analysis of variation in venous measurements implicitly includes any variation in parameters describing the shift of the oxygen dissociation curve, e.g. partial pressure of oxygen in the blood at SO2 of 50%, or 2,3-diphosphoglycerate concentration. Similar analysis has also been performed for variation in the values of RQ and ΔBEav 7. Variation in ΔBEav of 0.2 mmol·L−1 gives rise to errors in calculated pH, PCO2 and PO2 of 0.006, 0.08 kPa and 0.07 kPa, respectively. These errors are small, but it is clear that errors in ΔBEav amouting to several millimoles would be significant. The results presented in the current study and previously 8 are inconsistent with large changes in base excess across the peripheral sampling site even during sepsis. It is possible, however, that some caution is warranted in situations of acute changes in peripheral perfusion or acid–base status, such as hypovolaemic shock. Previous analysis of the sensitivity to variation in RQ has shown that the literature has reported a value of RQ variation equal to 0.08 15, which gives rise to small errors in calculated values of pH, PCO2 and PO2 equal to 0.005, 0.10 kPa and 0.06 kPa.
The largest source of measurement error is due to the pulse oximetry measurement of Sp,O2 (fig. 3⇑). Errors in Sp,O2 of ±4% gave rise to a small variation in calculated arterial pH that was fairly constant over the range of Sp,O2. A similar picture was seen in the error in calculation of arterial PCO2. The errors in calculation of arterial PO2 are a little more complex to interpret than for pH and PCO2, but some important conclusions can be drawn, as follows. From a clinical perspective, a pulse oximeter that reads too low, as represented by the bottom dashed line on figure 3c⇑, results in an underestimate of arterial PO2 and potentially more aggressive treatment of the patient with supplementary oxygen, fluids or other intervention. These interventions carry their own risk, but the risk of hypoxaemia associated with a pulse oximeter that reads too high is not present. For a pulse oximeter that overestimates oxygen saturation, the maximal error exists when the true value is 96% and the pulse oximeter reads 100%. As shown on figure 3c⇑, the error at this level is large but clinically unimportant, as both the true and overestimated PO2 levels are normal or above. In neither case would the patient require oxygen therapy.
The effects of error in Sp,O2 on the important clinical assessment, i.e. whether the patient has a low arterial PO2, can be seen in figure 3d⇑. If three situations where a pulse oximeter overestimated Sp,O2 are considered, labelled A, B, and C on figure 3d⇑, conclusions can be drawn as follows. From the difference in calculated and “true” arterial PO2 indicated by label A on figure 3d⇑, it can be seen that if the calculated PO2 is ≥12.5 kPa, then the true value is >9 kPa. From the point labelled B, it can be seen that if the calculated PO2 is ≥10 kPa, then the true value is >8 kPa. From the point labelled C, it can be seen that if the calculated PO2 is ≥8 kPa, then the true value is >7 kPa. This analysis shows that errors in Sp,O2 are unimportant to the calculation of arterial pH and PCO2. It also shows the potential of calculated values of PO2 to identify patients in need of supplementary oxygen, or where an arterial sample might be most beneficial. The improving precision of calculated values of PO2 at lower oxygenation shows that the error in Sp,O2 is least important in the most critical clinical situations. This is seen in both this sensitivity analysis, and from the results of the current study (fig. 2⇑), where the errors in calculated PO2 illustrated in the Bland–Altman plot were reduced with lower oxygen levels.
The venous–arterial calculation method assumes that the peripheral venous sampling site is from a limb with a clearly recognisable pulse and a normal capillary response, such that the anaerobic metabolism at the tissue site is negligible. The present study used routinely available blood samples and, thus, no attempt was made to ensure that this was the case, above that of normal clinical practice. It is, therefore, interesting that the results are in line with previous, experimentally controlled conditions.
The present study has evaluated a method for calculating arterial values of pH, carbon dioxide tension and oxygen tension from measurements in peripheral venous blood and a pulse oximeter, for patients presenting at a department of pulmonary medicine. It has been shown that values of arterial pH and carbon dioxide tension can be calculated precisely from peripheral venous blood and that values of arterial oxygen tension can be calculated with precision that may be clinically useful, at least when peripheral oxygen saturation ≤96%. The current study has shown the potential for use of the method in departments of pulmonary medicine, where it might replace routinely taken arterial samples, eliminating the need for painful arterial punctures.
Statement of interest
Statements of interest for S.E. Rees and M. Toftegaard, and the study itself, can be found at www.erj.ersjournals.com/misc/statements.dtl
- Received September 12, 2008.
- Accepted December 11, 2008.
- © ERS Journals Ltd