Automation in urinalysis: evaluation of three urine test strip analysers

A clinical laboratory evaluation was conducted on the Clinitek Auto 2000, the Super Aution Analyzer and the Urotron RL9 for the determination of glucose, protein, pH, blood, ketone-bodies and bilirubin. Precision of the systems was tested using three commercial control urine materials, and reported as the percentage of times the instrument repeats a certain value. Good repeatability was obtained with all the instruments. Accuracy of the systems was evaluated by comparison with quantitative procedures, and to check agreement between methods yielding semi-quantitative and quantitative results, ranges of acceptability were defined, based on the criteria reported in a previous paper [2]. It was then found that 87.5 to 98.9% of results from the Urotron RL9 and the Clinitek Auto 2000 were acceptable. With the Super Aution Analyzer the level of agreement was apparently lower because of the higher number of concentration steps used by this instrument.


Introduction
Urinalysis is now a very commonly performed, routine procedure and with the introduction of 'dipsticks' detection and measurement has become rapid and practical. More recently, the introduction of automatic analysers based on reflectance photometry has overcome the difficulty of subjective visual interpretation and allowed the achievement of better standardization.
The aim of this study was to investigate the performance of the most recently developed instruments: Clinitek Auto 2000 (Ames Co., Division of Miles Laboratory, Cavenago Brianza, Milano, Italy), Super Aution Analyzer SA4220 (Kyto Daiichi, Kogaku, represented in Italy by A. Menarini, Firenze) and Urotron RL9 (Boehringer Biochemia Robin, Milano, Italy).
The use of international guidelines for the evaluation of clinical chemistry analysers, designed for instruments providing quantitative results, gives rise to some difficulties when applied to instruments producing semiquantitative values [1 and 2]. Nevertheless, an endeavour was made to follow standardized protocols proposed for the evaluation ofanalysers in clinical chemistry [3 and 4] and their practicability was tested for analytical procedures leading to semiquantitative results.
In the evaluation, findings for urinary pH, protein, glucose and blood were compared with those from chosen quantitative methods. For bilirubin and ketone bodies appropriate quantitative methods were not easily available, so only recovery experiments were performed. Urobilinogen was not studied because there was no satisfactory reference material available nor was there a valid and practical comparison method. Leukocytes and nitrite were not taken into account.

Instruments
Characteristics of the three urine-analysers are given in Analyzer were calibrated every day. Urotron RL9 was calibrated every two weeks as recommended by the manufacturer, but the entire system was checked daily using an appropriate control strip (supplied by Boehringer).

Reagent strips
The Urotron RL9, Combur9Test RL strips have to be dipped manually into each urine sample following an optical and acoustic signal. The strips are then read automatically within min. The period of immersion of test strips in the specimen is standardized at 2 s. The Super Aution Analyzer employs a robot arm to provide complete automated operation for the entire procedure, from picking up the test strip to measurement. Reagent cassettes are used with the Clinitek Auto 2000 and contain a roll ofplastic film with 400 separate reagent areas attached. These are impregnated with chemicals for the determination of one type of analyte. Clinitek Auto 2000 performs urine-analysis by automatically aspirating and dispensing 100-200 [1 of the urine sample on each test pad.

Patients' specimens
Fresh urine samples submitted to the laboratory were selected for the study and stored at 4 C for a maximum of 4 h until analysed. To avoid interference no preservative was added to the samples.
It is well known that ascorbic acid may interfere with some tests on urine chemistry strips [5] and for this reason only fresh urine samples were selected after a negative ascorbic acid test by Rapignost Total Screen R (Istituto Behring, Scoppito, Aquila).
For the recovery of experiments, normal urine samples were collected from laboratory personnel. The specimens were negative for glucose, protein, ketones, bilirubin, 121

Recovery experiments
For erythrocyte detection and measurement, 0" ml of an EDTA treated whole blood sample containing 5"3 x 106 erythrocytes/l was added to 99"9 ml of 0"9% NaC1 solution. Different aliquots of this stock solution were diluted to 12 ml with normal urine (free of blood) to give 122 final concentrations between 5 and 600 erythrocytes/d. For reference a cell count was performed on an uncentrifuged urine using a Neubauer counting chamber.
For ketone-bodies and bilirubin artificially supplemented urines were prepared by adding pure substances. Results obtained from the automatic urine-analysers were compared with the theoretical concentrations.
Results and discussion Imprecision Ten determinations were performed with each urine control material on five consecutive days. The results are summarized in tables 2-4.
Standard deviations and coefficients of variation cannot be calculated for test strip analysers operating in discrete mode, so in the tables the percentage of times the instrument repeats a certain value is indicated. Good repeatability is reported as 100% when the same value is always obtained. A degree of imprecision may depend on analyte concentration being at the border between two concentration ranges.
For this reason, with these instruments, a variation of one step lower or higher than the expected one [8] is generally accepted to be a reasonably good precision.
Anomalies were found using experimental control material, particularly with the Super Aution Analyzer and with the Urotron RL9 for bilirubin in Kova-Trol I (see table 4). This variability can be largely attributed to the characteristics of the testing materials. Good repeatability was observed with patients' samples. In the absence of a precise indication by the manufacturer, the borderline values of a concentration range, were set exactly half-way between two adjacent set points [9 and 10]. As in a previous paper [2] agreement between the various analysers and a quantitative comparison method was considered good if the value obtained with the comparison method did differ from the concentration range of a set point, plus or minus 50% of the adjacent ranges.  2[a]).
With the Urotron RL9 good results were determined for glucoseno outlier ( figure 2[b]) and proteins-4% out of acceptable range ( figure l[b]). Results from pH ( figure  3[b]) were poor (12'5% not acceptable), despite the concentration ranges being double those of the other instruments.
The results from the Super Aution Analyzer need to be carefully evaluated; higher percentages of values outside the acceptable range (37"5% for glucose, 50"5% for protein and27" 1% for pH) are mainly due to the greater number of concentration intervals which is a feature of this instrument. Since precision is similar to the other two instruments, it is suggested that the number of intervals should be made the same. With the instrument having a higher number of intervals (and, consequently, very small ranges of concentration), the increased number of results outside the acceptable intervals is understandable. Figure 4 shows the results obtained when different concentrations of erythrocytes were added to normal urines. With the Urotron RL9, 92"2% of results agreed with the comparison method. With the Super Aution Analyzer, 41"2% of results were underestimated. This    was probably due to the sensitivity of the test pad being lower for erythrocytes than for free haemoglobin. The Clinitek Auto 2000 provides only qualitative results for haemoglobin content so acceptable ranges were produced by arbitrarily assigning semiquantitative values to qualitative expressions; 92% of results fell into the assigned ranges.

Practicability
The operation of all the instruments was simple; maintenance required only a few minutes a day and no malfunctions were observed during the evaluation period which lasted about six months. The volume of specimen needed for the three urine-analysers differed. With the Urotron and Super Aution test strip dipping requires at least 12 ml; the minimal volume required for the Clinitek was 2 ml.
the results falling in adjacent ranges must be accepted [2]. Therefore, when the instrument reports for example, 100 mg/dl of glucose, the real value could be between 60 or 190 mg/dl ( figure 2[b]). When the number of intervals is increased in an effort to reduce such an uncertainty, as for the Super Aution Analyzer, there appears to be greater inaccuracy. Figures 1-4  In conclusion, no major differences appear to exist in the analytical quality of the three analysers; they represent the state of art of urine analysis but an improvement in accuracy is desirable.,