Assessing the ABL 500 blood gas analyser

The ABL 500 blood gas analyser from Radiometer has cordless electrodes and does not use a humidifier for calibrating gases. During the evaluation of the analytical performance of this instrument, the problem of p02 accuracy was approached by comparing the values obtained with two kinds of tonometry (film and bubble). An acceptable level of imprecision was demonstrated for all measured parameters. For within-run precision, with tonometry, coefficients of variation (CV) were ≆0.37% for pO2 and ≆0.52% for pCO2. A CV of 1.76% was found for day-to-day precision for both p02 and pCO2. In the linearity study, with both tonometry methods, and in the inter-instrument comparisons (the ABL was compared with the Ciba Corning 178), pO2 values obtained on the ABL 500 exhibited a slight overestimation above 150 mmHg (2.2-3.4% at 600 mmHg). This minor discrepancy is discussed with reference to the new design of the pO2 electrode, the algorithm for pO2 correction and the tonometry procedure. The results reported in this paper stress the importance of pO2 accuracy assessment for the evaluation of blood gas analysers.

The ABL 500 blood gas analyser from Radiometer has cordless electrodes and does not use a humidifier for calibrating gases. During the evaluation of the analytical performance of this instrument, the problem of p02 accuracy was approached by comparing the values obtained with two kinds of tonometry (film and bubble). An acceptable level of imprecision was demonstrated for all measured parameters. For within-run precision, with tonometry, coefficients of variation (CV) were <0"37% for pO2 and <0"52% for pCO. A CV of 1"76% was foundfor day-to-@ precision for both p02 and pCO2. In the linearity study, with both tonometry methods, and in the inter-instrument comparisons (the ABL was compared with the Ciba Corning 178), pO values obtained on the ABL 500 exhibited a slight overestimation above 150 mmHg (2"2-3"4% at 600 mmHg). This minor discrepancy is discussed with reference to the new design of the pO electrode, the algorithm for pO correction and the tonometry procedure. The results reported in this paper stress the importance of pO accuracy assessment for the evaluation of blood gas analysers.
following gas mixtures from atmospheric air and from pure CO2: GI: 5"6% CO2, 19"76% 02; G2: 10% CO2 is the high point of calibration of the pCO2 electrode; G3 is 100% CO2 and is used for the zero calibration of the pO2 electrode. The ever-present film of rinse solution inside the tubing is sufficient to achieve the necessary humidification of the calibration gases. The pH electrode is calibrated using two reagent buffers. Automatic calibrations are adjusted according to individual requirements. The ABL 500 accepts two sample sizes: 35 1 for pH in a special micromode; a total blood gas analysis is available from 70 gl using injection or an aspiration mode.
The Ciba Corning 178 blood gas analyser was used as comparison instrument.

Solutions
Phosphate buffer solutions at pH 6"838 and 7"384 were prepared in line with the recommendations of National Bureau of Standards [5].
The Radiometer ABL 500 blood gas analyser is modular and includes some highly innovative features, such as cordless electrodes and a remembraning system. The analyser was evaluated for analytical performance and practicability, using both tonometry and commercial aqueous control solutions [1]. A recurrent problem in evaluation in blood gas instrumentation is pO2 accuracy [2] built-in corrections apply to the directly measured pO2 values. The accuracy of pO2 displayed by the instrument depends upon the quality of the algorithms used for correction. Qualicheck Radiometer aqueous control solutions were used at three levels (acidosis, normal and alkalosis) and a fourth high oxygen level was used for the within-run study. The equation needed for this adjustment is determined empirically during the development of a new instrument by comparing data from several uncorrected analysers to a reference method [3]. There is no standard method for determining the partial pressures of blood gases with absolute accuracy [4]. However, tonometry of fresh whole blood is considered the reference method for assessing pO2 accuracy. In order to test the validity of the ABL 500 algorithm for pO2 correction, two kinds of tonometry were used (film and bubble) and the reliability ofpO2 measurement was assessed according to the results obtained by the two methods.

Materials and methods
The Laue bulb tonometer consists of a rotating glass bulb, placed in a water-bath at 37.0C with a gas humidifier and a gas vent. Due to the excentric rotation of the bulb a thin film of the sample is formed. The Corning 184 tonometer is designed as a syringe tonometer according to the bubble equilibration principle. Antifoam solution was added to blood samples (Corning antifoam solution). Both tonometers were operated and maintained according to the manufacturer's instruments.

Blood samples
Fresh heparinized venous or arterial blood was obtained from healthy donors for tonometry and from hospitalized patients for comparison studies.

Blood gas analysers
The ABL 500 measures pH, pCO and pO2. Electrodes are based upon reference technology with miniaturized cordless electrodes, which are colour coded to ensure correct placement. The built-in gas mixer supplies the Protocol pH accuracy The two phosphate buffers were run in triplicate over five days. Day-to-@ precision The same solutions and same levels of tonometry were run daily for 25 days.
Drift was assessed using tonometered blood tested immediately after one calibration and before the subsequent one.
Linearity was tested using successive measurements of tonometered blood containing O2 for 0 to 85% and CO2 from to 20%. The sequence was repeated three times with the two systems of tonometry.

Inter-instrument comparisons
About 300 samples from patients were simultaneously measured on both instruments. The values covered the patho-physiological ranges for the three parameters under investigation.

Practicability
Special attention was paid to quantifying the most important specifications of the instrument, to maintenance requirements and to safety.

Results
The mean pH values from the triplicate measurements were never separated by +0"01 UpH from the assigned values of the two buffers.

Precision study
Within run precision Coefficients of variation never exceeded 0.43% for pCO2 and 0"73% for pO2 at each level of Qualicheck solutions.
With film tonometry CVs were <0"52% for pCO2 and <0"37% for pO2. Results for a representative sequence of the within run precision study are given in table 1.
Results for day-to-day precision are given in table 2 and illustrate that the measured values by tonometry were very close to the assigned values.
No drift was observed between calibrations.

Practicability
The ABL 500 is based on a modular design with a wet section located in the front part of the measuring station, sensitive electronics in an independent module, and a small CO2 cylinder inside a cabinet frame, saving space and costs. The measuring chamber is clearly visible. Remembraning time is reduced with disposable membrane units, which are prefilled with electrolytes solutions. The PC-style software is ingenious and easily accessible (finger wheel, pop-up windows etc.). Maintenance is guided by an adaptable self-diagnostic program. The mechanical steps are simple and easily learnt. Reagents and waste containers are designed to conform to strict biological safety rules. The inlet section provides easy access for sample introduction and cleaning of the protective flap. The automatic results print-out is easily edited and items can be selected from it.

Discussion
The evaluation demonstrated a high degree of precision for all measured parameters. Inter-instrument comparisons and reference method application identified some discrepancies especially for high pO2 values.
The linearity study gave the following results: (1) The zero point calibration was verified by tonometry.
The measurement of the electrode response, at zero pO2, increases the accuracy of the subsequent determinations.
(2) Near the second calibration point of the slope (02 20%), the measured values were identical to the expected ones.
The values given by the analyser are always corrected values. The algorithm for correction is dependent upon the gas/liquid ratio and the sensitivity of the pO2 electrode, calculated during the last calibration of the high pO2 gas. There are also corrections for systematic deviations arising from the contamination of the sample by the amount of oxygen present on the inner walls of the tubing and/or measuring chamber. The magnitude of this effect is influenced by different factors, such as the .__ ratio between sample volume and contact surface, the oxygen buffer capacity of the sample and the surface layer of the walls and finally the time of contact between sample and wall. Although tonometered, fresh blood remains the best means for testing instrument accuracy. It requires a strict mmHg. (y l'04x 2"71, r 0"998).
control of working conditions: the IFCC method was followed for tonometry of blood, particularly in terms of procedure and equilibrating conditions [6]. Improper handling during the anaerobic use of the Laue tonometer and gas humidification, for instance, yield inaccurate values [7]. Even at gas flow rates of 60 ml/min and (y 1"03x + 1"94, r 0"999). equilibrating temperature maintained at 37"0 + 0" 10 C could not account for the differences observed between the assigned and the measured values at high PO2 levels.
In the Corning 184 tonometer, the blood sample is transferred directly from the equilibration syringe to the blood gas instrument, but changes in the sample may have several explanations: (a) Haemolysis due to the antifoam solution, but the consequence of this is minor magnitude [8]. (b) A bubble effect, which raises pO2. This bubble effect is related to the surface tension of the liquid surrounding the bubble, the bubble diameter and the hydrostatic pressure in the tonometer vessel [9].
The difference in pO2 between the Corning 184 and the Laue tonometer at a level of 600 mmHg (80 kPa) was 7"5 mmHg (1 kPa) and was due to the bubble effect. An estimation of this bubble effect has previously been calculated at 13"5 mmHg (1.8 kPa) for pO at 675 mmHg (90 kPa) between the Corning 184 and IL 237 tonometers [10]. However, beside the discrepancies demonstrated by tonometry, an overestimation of high pO values persists on the ABL 500 analyser. For example, the measuring time on blood samples is not constant for the pO2 electrode. The value calculated from the second reading is used to determine the measuring time. When pO2 is very low or very high,, this time is maximal. In addition, the inner side of the polypropylene membrane is covered by platinum black, resulting in a faster and more stable pO electrode. Theoretically, these new features should have improved the stability and the accuracy of the pO electrode.
The algorithm was established using capillary tubes and aspiration mode, as on the previous models of analysers by Radiometer 11]. Under these conditions the volume of the sample and the magnitude of the contamination were precisely known. However, for the present evaluation, syringes and the injection mode were used. The volume of the sample is probably larger and more variable and the contamination not identical to the conditions defined by the aspiration mode. However, the injection mode with a syringe is more generally use?t for routine measurements. Moreover, pO values were not identical above 150 mmHg (20 kPa) between the Ciba Corning 178 and the ABL 500, when comparing patients' samples. These two facts suggest that the small overestimations of the ABL 500 pO determinations at elevated oxygen pressures are of minor importance. Difficulties in handling samples with very high oxygen tensions must be noted. Most performance specifications given by the manufacturers do not reach pO2 values above 550 mmHg. To further improve pO accuracy, the algorithm for correction should take into account high pO tensions and the mode of sample introduction.
In conclusion, this study emphasizes that assessment of pO2 accuracy is still difficult. The quality control materials which are available today are not ideal.
However, commercially prepared materials are appropriate on a routine basis [12]. Tonometry is still questionable, but it is the only way to assess the characteristic properties of the blood gas analysers: imprecision, inaccuracy and inter-instrument variations [13]. The slight inaccuracy observed for hyperoxic levels on the ABL 500 analyser should be balanced by the reduced clinical interest in the pO2 reliability in this high range. The technical innovations associated with the up-to-date computer style of this analyser makes it particularly easy to run after a short period of training.