I

CLIN. CHEM. 31/3, 435-438 (1985)

The Technicon RA-1000 Evaluated for Measuring Sodium, Potassium,

Chloride, and Carbon Dioxide
Ronald H. Ng, Marcia Aitaffer, Ralph fto, and Bernard E. Statiand

We evaluated the Technicon RA-1 000#{174} random-access sensitive electrode and a silver/silver chloride reference
analyzer for the measurements of sodium, potassium, and electrode. For CO2 measurements, a carbonate-sensitive
carbon dioxide by an indirect potentiometnc method (ion- electrode (2,3) is used in the RA-1000 and a Pco2 electrode
selective electrode) and for chloride by a colorimetric method (measuring the rate of change of the pH differential) is used
(mercuric thiocyanate). For various concentrations of control in the ASTRA-8. The RA-1000 ISE module has three basic
materials the total precision (CV) rangedfrom0.9 to 1.2% for components: ISE assembly, sample assembly, and punip/e-
sodium, 1.1 to 1.3% for potassium, 1.0 to 1.2% for chloride, lectronic assembly. After dilution with a buffer in the
and 2.8 to 3.8% for carbon dioxide.The systemdemonstrat- sample assembly, the sample is aspirated through the ISE
assembly for analysis simultaneously with a counterfiow
ed acceptable performance in linearity and carryover. Pa-
reference reagent solution. The pump/electronic assembly
tients results from the RA-1000 correlated well with those
controls the movement of sample and reagent through the
from the Beckman ASTRA-8#{174}. In a study on potential inter-
ISE assembly and the electronic signals between the assem-
ferences, we found that high concentrations of salicylate and
blies. The RA-1000 is calibrated with a two-point calibra-
bromide significantly affected measurements of carbon diox-
tion, i.e., with high- and low-concentration calibrators. At
ide and chloride, respectively. The RA-1 000 requires only 30 the beginning of each worklist (run) and approximately
p.L of sample for all four tests and it offers a high throughput every 8 mm during the course of analysis, the ISE automati-
(30 specimens analyzed for the four tests in 25 mm). This cally analyzes a wash solution as a one-point calibration, to
precise, easy-to-use, random-access analyzer requires mini- adjust the baseline if necessary and to check for drift, and
mal maintenance. the voltage range.
To measure chloride in the RA-b000, a colorimetric meth-
AddItIonal Keyphrases: random-accessanalysis . ion-selective od with mercuric thiocyanate (4) is used; in the ASTRA-8 a
electrodes electrolyteprofile chloride electrode (coulometric-amperometric titration) is
used. In the HA-bOO chloride method-an equilibrium,
Ion-selective electrodes (ISEs) have gained increasing bichromatic assay-the assay mixture is measured at 500
popularity in recent years since simple, precise, and easy-to- nm after 3 mm of reaction; bichromatic blanking is made
use ISEs have become available. With a recently introduced with measurements at 600 nm.
ISE module, the RA-1000 analyzer (Technicon Instruments Calibration: The HA-bOO ISE was calibrated every 4 h
Corp., Tarrytown, NY 10591) now measures sodium, potas- with two solutions: a high calibrator (containing 159
sium, and carbon dioxide in addition to performing chemical mmol of Na, 10.3 mmol of K, and 48.0 mmol of CO2 per
and immunoassays. A bench-top, computer-controlled ana- liter) and a low calibrator (with respective concentrations
lyzer, the HA-bOO can perform, without batching, single or of 96, 2.2, and 12.0 mmol/L). We calibrated C1 every two
multiple tests in any combination or sequence (1). We weeks with a serum-based material containing 102 mmol of
evaluated the analytical performance of the RA-1000 ISE Cl per liter. The ASTRA-8 was calibrated every 75 mm
for Nat, K, and C02, and the colorimetric assay for with two serum-based calibrators: a high calibrator (per
chloride (Cl), together constituting the most common elec- liter, 172.2 mmol of Na, 8.8 mmol of K, 125.4 mmol of
trolyte profile in clinical laboratories. We attempted to Cl, and 31.24 mmol of C02) and a low calibrator containing
investigate all potential limitations and interferences and the respective analytes at 118, 3.7, 52.5, and 10.35 mmol/L;
report them as well as any favorable findings of the system. both contained ethylene glycol as preservative. After cali-
bration, we monitored the stability of RA-b000 performance
Materials and Methods by measuring the control or calibrator material both more
frequently than and also beyond the time recommended by
Instruments. Na, K, Cl and CO2 were analyzed with the manufacturer for a new calibration.
both the RA-1000 and the ASTRA-8 (Beckman Instruments, Precision. Precision was assessed according to the Nation-
Inc., Brea, CA 92621). Both systems involve a glass sodium- al Committee for Clinical Laboratory Standards (NCCLS)
sensitive electrode, a valinomycin-membrane potassium- protocol EP5-P by assaying quality-control materials and
plasma pools in duplicate in the morning and in the
afternoon for 11 days. The total SD and CV were calculated
Department of Laboratory Medicine, University Hospital, Boston from all results n each of three levels of quality-control
University Medical Center, Boston, MA 02118.
Presented in part at the 36th national meeting of the AACC, materials.
Washington, DC, July, 1984. Linearity. Technicon TQC Alert#{174} 1 and 2 lyophilized
Received October 9, 1984; accepted December 6, 1984. serum-based quality-control materials were reconstituted in

CLINICALCHEMISTRY,Vol. 31, No. 3, 1985 435

various amounts of de-ionized water to prepare each in a The RA-1000 responses varied linearly with electrolyte
high and a low concentration. Equal volumes of these two concentration (Table 2).
preparations were mixed to produce a middle concentration; Correlation of results for plasma samples as measured
we also prepared mid-high and mid-low concentrations. with both the RA-b000 and the ASTRA-8 is shown in
Nat, K, and Cl measured in these preparations were Figures 1-4. Sodium in plasma as measured with the HA-
compared with the calculated expected values, and linearity 1000 averaged 1.4 mmol/L (1%) lower than with the AS-
was assessed by linear regression analysis. The TQC Alert TRA-8, K values averaged 0.25 mmol/L (6%) lower, C1
control materials are not recommended for routine use as an values averaged 0.05 mmol/L (<0.1%) higher, and CO2
HA-bOO ISE control because they contain enough salicylate values averaged 0.4 mmol/L (1%) higher.
to interfere with CO2 measurement. Thus, to assess the Technicon Alert-2 control was analyzed for Na and K
linear range for C02, we used a serum-based quality-control in quadruplicate immediately and 2, 4, 6, and 8 h after a
material reconstituted with solutions of sodium bicarbonate
in various concentrations.
Method comparison. Using both instruments, we analyzed Table 2. LinearIty Study
more than 100 patients plasma samples according to the Regr.ulori analysis (n =5 each)
Concn range,
NCCLS protocol PSEP-4. Each specimen was analyzed in mmol/L Slop. Intercept r ,
duplicate for Na, K, and C1 and in singlet for CO2 on Na 50-200 1.00 0.4 1.000 0.6
both instruments within a 2-h time. No more than 20 K 1.3-9.8 1.00 -0.2 0.998 0.2
specimens were analyzed on a single day. Outliers were C1 80-130 0.99 2.5 0.998 1.5
detected and eliminated, when present, from regression CO2 12-50 1.02 -1.7 0.999 0.5
analysis according to the NCCLS protocol.
Carryover. Technicon TQC Alert control was reconstitut-
ed in de-ionized water and salt solution to serve as the low RA-1000 Values = 0.955 Astra Values + 4.6
pool and the high pool, respectively. The salt solutions for r0.972
the measurements of Na, K, C1, and CO2 were NaCl Sy.x = 1.6
1500 mmol/L, KC1 60 mmol/L, NaCl 1000 mmol/L, and 160- n102
NaHCO3 300 mmol/L, respectively. Thus, the high pool -J .
was a serum-based material with 10 times the physiological N
0 . .
concentrations of Na, K, C1, and CO2. Four samples of
the high pool were followed by four samples from the low E #{149} #{149}#{149}
E 140
pool in the assay. The difference in concentration between
the first low pool sample and the average of the last two E
low pool samples was divided by the concentration of the
high pool as an estimate of the amount of carryover.
Interferences. Technicon TQC Alert control was reconsti-
.5

0
0
Co
120
I
S..
#{149}
tuted in solutions containing high concentrations of lithium, 0
0
calcium, ammonia, magnesium, and bromide. The Na and
K results for these preparations were compared with those
for the same control material reconstituted in de-ionized
water. To study salicylate interference with the CO2 elec- I i i
trode, we used patients specimens known to contain high 100 120 140 160
concentrations of the drug; we also added salicylate to drug- ASTRA-8 Sodium mmol/L
free pooled plasma. Concentrations of ammonia and saucy-
late were determined with an aca discrete analyzer (DuPont Ag. 1. Sodium in plasma as measured with the ASTRA-8 and the RA-
Instruments, Wilmington, DE 19898). To study interference 1000
from hemolysis and icterus, we determined the analytical
recoveries of Na, K, Cl-, and CO2 after adding standard RA-1 000 Values = 0.936 Astra Values +0.005
= 0.996
solutions of NaC1, KC1, or NaHCO3 to icteric or hemolyzed 9 Sy.x = 0.08
plasma pools. To evaluate the effect of lipemia, we analyzed n= 102
lipemic plasma or serum samples before and after ultracen-
trifugation. -l

o
Results E
E
For various concentrationsof quality-control materials,
the total CVs (11 days, 44 analyses) were no more than 1.3% E
..
#{149}
for Na, K, and C1, and no more than 3.8% for CO2 (Table a
1).
4
Table 1. PrecisIon of RA-1000 In Measuring Three
Concentrations of Electrolytes
Mean, Mean, Mean,

Na
K
Cl
CO2
mmol/L
115
3.2
89
13
CV, %
0.9
1.3
1.2
3.8
mmol/L
137
5.8
104
19
CV, %
1.3
1.2
1.0
2.8
mmol/L
148
8.9
115
25
CV, %
1.2
1.1
1.0
3.3
Il ASTRA-8 Potassium. mmol/L
n= 44 each.
Fig. 2. Same, for potassium in plasma

436 CLINICAL CHEMISTRY, Vol. 31, No. 3, 1985

 RA-1000 Values 0.993 Astra Values + 0.8 The amounts of carryover detected in this study were
r = 0.986 0.48% of 1538 mmol/L, 0.32% of 63 mmolfL, 0% of 1010
140 Sy.x = 2.0 mmolJL, and 0.56% of 444 minol/L for Na, K, Cl, and
CO2 measurements, respectively.
n = 127 In the interference study, we found no more than 2
-I
mmol/L (1.6% of 125 mmolfL) and 0.1 mmol/L (1.8% of 5.4
0 C.
E 120
mmol/L) variations for Na and K, respectively, in the
E presence of 5 mmol/L lithium, 10 mmol/L calcium, 2 mmol/L
0 ammomum chloride, or 5 mmol/L magnesium. Ammonium
sulfate at 2 mmol/L did not interfere with CO2 measure-
0 ments. Bromide at 10 and 20 mmol/L produced positive
C.) ;1#{149} biases of 20 and 42 mmol/L, respectively, for chloride
100
0 measurements, but did not interfere with Na, K, or CO2.
0
0 Salicylate was found to interfere with the CO2 measure-
ment. An average bias of +4 mmol/L was found at 150 mgfL
(1.08 mmol/L) salicylate. The effect of salicylate on the CO2
80 electrode is short-lived and reversible. It affects the salicy-
late-containing specimen and no more than three consecu-
.1.
80 100 io io tive specimens that follow.
No significant interference was found with moderately
icteric or hemolyzed specimens. Analytical recoveries of
ASTRA-8 Chloride, mmol/L standard solutions added to the abnormal plasma or serum
Fig. 3. Same, for chloride in plasma specimens were within 93 and 102% in the presence of 200
mg of bilirubin or 10 g of hemoglobin per liter.
RA-1000 Values = 1.016 Astra Values - 0.001 Moderate lipemia (triglyceride of 7 g/L and an absorbance
r = 0.966 of 0.5 at 600 nm and 1-cm light path after 10-fold dilution in
water) produced a positive bias of 5 to 7 mmol/L in Cl
50 Sy.x = 1.09
measurement but had no effect on Na, K, or CO2. Severe
n = 441 lipemia produced a negative bias in Na values and a
positive bias in CO2 values. For example, a grossly lipemic
specimen (triglyceride of 33 g/L) caused a negative bias of 5
mmolfL for Na with the HA-bOO and 8 mmol/L for Na
with the ASTRA-8. The HA-bOO detected all positive
interference with CO2 and Cl- caused by severe lipemia and
printed an error code next to the result. Severe lipemia is
detected by monitoring the change in absorbance at 600 nm
before and after sample addition in the C1 assay, and by
the electrode response curve in the CO2 assay. Serum
blanking is not programmable with the Cl assay, which is
a bichromatic method.
Actual analysis
time for measurements of Na, K, Cl-,
and CO2 in 30 specimens was 25 mm with the RA-1000, 31
miii with the ASTRA-8.

Discussion
ASTRA-8 Carbon Dioxide, mmoi/L
We have evaluated measurement of Na, K, C1, and
Fig. 4. Same, forcarbon dioxide in plasma
CO2 in the HA-bOO, with particular attention to potential
technical problems that may affect results. The precision
calibration. The mean values at different time intervals study conducted according to NCCLS protocol demonstrated
varied by no more than 1 mmol/L and 0.08 mmol/L for Na satisfactory performances for all four tests. The CVs were
and K, respectively (Table 3). For CO2, the mean values (n comparable to those exhibited by the ASTRA-8 system in
= 4) of the high calibrator varied by no more than 2 mmol/L our laboratory.
up to 8 h after a calibration. For C1, the mean values (n = Acceptable performance in linearity was confirmed with
4) of the quality-control serum varied by no more than 3 regression analysis of our data for the four assays. The
mmol/L for as long as four weeks after a calibration. The linear ranges are sufficient for clinical need.
HA-bOO rarely failed in calibration throughout our evalua- For method comparison, we compared many HA-bOO
tion. results for patients specimens with those obtained with the
ASTRA-8, finding a good correlation.
Table 3. RA-1000 Performance after Calibration Although the manufacturer recommends calibrating the
Mean (and SD), mmol/L ISE every 4 h, we found that a calibration can hold up to 8 h
Tim., Time, Cr, mean for Na, K, and CO2 measurements. For chloride, we found
h Ns CO2 days (and SD), mmol/L that the calibration curve was stable for at least four weeks.
0 125.7 (0.5) 5.47 (0.05) 50.5 (0.9) 0 113.2(0.5) Compared with the required calibration of the ASTRA,
2 124.7 (1.2) 5.45 (0.06) 51.0 (1.4) 7 112.2 (1.4) every 75 mm, the much less frequent calibrations needed
4 126.0 (0) 5.42 (0.05) 51.75 (0.5) 14 110.2 (0.5) with the HA-bOO save reagent, calibrator, and quality-
6 126.0 (1.1) 5.55 (0.06) 52.0 (0) 21 112.5(0.6)
8 126.0 (0) 5.50 (0) 51.75 (0.5) 28 113.2 (0.9) control materials as well as instrument time. This feature
becomes more important when rapid turnaround of results
n= 4 each.
is needed on stat requests and a calibration cycle (e.g.,

CLINICALCHEMISTRY,Vol. 31, No. 3, 1985 437

 approximately 7 miii with the ASTRA-8 for the four ana-
lyte8) can delay the process.
We found that the HA-b 000 had higher throughput than
the ASTRA-8 in measuring all four analytes. If Cl- is not
requested, the RA-1000 would be even faster, because it
takes as much time to analyze for Cl- as to do all the three
ISE tests, i.e., Na, K, and CO2.
Using a serum-based material with Nat, K, C1, and
CO2 added to 10-fold the physiological concentrations, we
detected negligible carryover with the ISE tests and con-
cluded that the system is satisfactory for clinical use. For
Cl-, which is a non-ISE assay in which random access
fluid is used in all sampling probes to encapsulate the
samples and reagents, no carryover could be detected.
In the interference study, high concentrations of cations
such as Li, Ca2, and Mg did not interfere with measure-
ment of Na and K. In patients with hepatic coma (5) or
severe Reyes syndrome (6), serum ammonia values as high
as 0.6 mmol/L have been reported. We showed that 2.0
mmol of ammonia per liter did not interfere with the HA-
1000 ISE for Na, K, or CO2 measurements.
Although bromide has been replaced by barbiturates
nowadaysin medical treatment, the former reportedly pro-
duces a positive interference with ISE and colorimetric
thiocyanate methods for Cl determination (7). We con-
finned this in the HA-bOO system, in which the thiocyanate
method is used. Thus, if a substantially supranormal chlo-
ride concentration is found in a specimen, it should be
confirmed coulometrically; if that method yields a signifi-
cantly lower result, interference by bromide should be
suspected. Nevertheless, bromide should seldom be a prob-
lem, because it is rarely used as a therapeutic agent today.
Because salicylate reportedly produces a positive interfer-
ence with the carbonate ionophore method for CO2 measure-
ment on the Kodak Ektachem 400#{174} system (8), we decided
to investigate its effect on the HA-bOO ISE. Indeed, we
detected a positive interference of 4 to 8 minol/L with CO2
measurement in patients sera that contained b50 to 300 mg
(1.08-2.17 mmol) of salicylate per liter. This was caused by
an interaction of salicylate with carbonate ionophore to
simulate high CO2 readings and shorten electrode response
time. Since then the manufacturer has revised the software
program on the HA-bOO so that it can detect any salicylate
interference exceeding approximately 4 mmol/L, and an
error code is printed (flagged) next to the CO2 result. The
revised software can distinguish an abnormal curve of CO2
electrode response that does not plateau within the analysis
time (5.5 s), as in the case of salicylate interference. We have
confirmed the ability of the new software (RA-b000 Program
Disk, version 2.3) to detect the salicylate interference. We
(and the manufacturer) recommend that the four samples
following a flagged sample be reanalyzed to veri1 proper 5. Durst HA. Automated analyzer for the determination of potassi-
CO2 values. If the repeated results of the four specimens
differ substantially from their original values, the rest of the
samples in the original run, excluding the flagged speci-
men, should be repeated. The flagged specimen should be
analyzed by an alternative method. In our evaluation,
exposure to salicylate did not appear to shorten the useful
life of the CO2 membrane.
Thus, unlike the Ektachem 400, the HA-b000 is pro-
grammed to limit undetected interference of salicylate, if
present, to an amount corresponding to no more than 4
mmol of CO2 per liter. In our method-comparison study, we
438 CLINICAL CHEMISTRY, Vol.31,No. 3,1985
randomly selected 441 patients specimens; none of the CO2
results were flagged with the error code, and no apparent
positive bias was found when the results were compared
with those of the ASTRA-8. We did our interference study
by screening many specimens from arthritis patients (who
may be taking a high dose of aspirin) and found a few
specimens with salicylate concentrations in the range 150-
300 mgIL. The dosage of salicylate resulting in concentra-
tions of 150-300 mg/L in plasma is generally 3-6 g per day
(e.g., 15 to 30 tablets of Excedrin, each containing 0.2 g of
salicylate) (9). In our hospital, we estimate that less than
0.1% of the specimens we receive have salicylate concentra-
tions >150 mgIL. Nevertheless, many commercial quality-
control products contain salicylate, so it is essential that
laboratories confirm the absence of salicylate in their con-
trols before using them in the HA-bOO ISE.
Moderate icterus and hemolysis did not significantly
interfere. Although the RA-b000 ISE is an indirect potentio-
metric method, only severe lipemia caused a significant
negative bias on Na results. The negative interference in
the HA-bOO was smaller in magnitude than that in the
ASTRA-8 because the effect of severe lipemia is diminished
when a lower dilution ratio is used: the RA-1000 dilutes the
specimen with 14 volumes of buffer, the ASTRA-8 with 26
volumes. Nevertheless, all severely lipemic specimens
should be analyzed by direct ISE or ultracentrifuged before
analysis by indirect ISE.
In summary, the interferences revealed by our study on
the RA-1000 were either insignificant or involved rarely
encountered specimens with easily detectable and correct-
able types of interferences. Thus, we believe that the HA-
1000, like the ASTRA-8 and Ektachem 400, is acceptable for
clinical use in measuring Nat, K, Cl, and CO2. We found
the HA-bOO to be precise, easy to use, and to require
minimal maintenance. The system also provides high
throughput and random-access capability.
References
1. Schwartz MK, Statland BE, Coughlin J, et al. Chemical and 4
clinical evaluation of the random access analyzer HA-bOO. Clin
C/tern 30, 364-368 (1984).
2. Diebler H, Adler H, Svenjak D, et al. A new approach for the
rapid, simultaneous determination of Na, K and total CO2 on the
Technicon RA-1000 system. Clin Chem 29, 1193 (1983). Abstract.
3. Scott W, Chapoteau E, Jensen M, et a!. Carbonate ion-selective
membrane electrode with improved characteristics. Clin C/tern 30,
966 (1984). Abstract.
4. Schoenfeld RG, Lewellen (11. A colorimetric method for deterrni-
nation of serum chloride. Clin C/tern 10, 533-539 (1964).
urnand sodium in whole blood. Clin Chim Acta 80,225-234(1977).
6. DeVivo DC, Keating JP. Reyes syndrome. Adu Pediatr 22, 175-
229 (1976).
7. Elm RJ, Robertson EA, Johnson E. Bromide interferes with
determination of chloride by each of four methods. Clin Chem 27,
778-779 (1981).
8. Steelman M, Smith CH, Menon A, et al. Interferences with
potentiometry of CO2 in the Ektachem 400 analyzer. Clin C/tern 30,
562-565 (1984).
9. Gilliland BC, Mannik M. Rheumatoid arthritis. In Harrisons
Principles of Internal Medicine, KJ Isselbacher et al., Eds.,
McGraw-Hill Book Co., New York, NY, 1980, pp 1872-1880.