A systematic evaluation of the Olympus AU5061 as an effective replacement for the SMAC II analyser

An instrument evaluation ofthe Olympus A U5061 was conducted by a National Committee on Clinical Laboratory Standards (NCCLS, 771 East Lancaster Avenue, Villanova, Pennsylvania 19085, USA) protocol. Reagents having the same lot number were obtainedfrom Data Medical Associates, Inc. (2016 Easl Randol Mill Road, Arlington, Texas 76011, USA). Formulations employed were those generally accepted as standard clinical chemistry methods. Control materials from Dade (American Hospital Supply Corporation, P. O. Box 520672, Miami, Florida 33152, USA) were analysed to determine within-run and day-to-day precision. Within-run precision CV) was in the range of0"31"4"6% for all methods andjudged to be better than that of the SMAC H Technicon Instruments, Tarrytown, New York 10591, USA). Linear ranges were equal or exceeded those available

unassayed lyophilized serum control materials were used for all precision measurements. Calibrators A 145/5"0 mmol/1 aqueous standard containing lithium (3 mmol/1) as an internal standard was supplied by Olympus and used to calibrate the flame photometer. Set Point (Lot number V6B258) and Set Point 2 (Lot number V6C261) from Technicon (Technicon Instruments Corporation, 511 Benedict Avenue, Tarrytown, New York 10591, USA) and the values supplied by Technicon were used to calibrate all other methods.
New England Reagent Laboratory (NERL) Standards Weighed-in standards obtained from NERL (14 Almeida Avenue, East Providence, Rhode Island 02914, USA) were used to measure linearity for all chemistry procedures except the enzymes. For the latter, Multi-Enzyme Lin-Trol (PN M2266) was obtained from Sigma (Sigma Chemical Company, P.O. Box 14508, St Louis, Missouri 63171, USA) and diluted to obtain multiple points. For bilirubin, cholesterol, triglycerides, GGT and CK dilutions of a high patient serum were used to measure the linear response of these methods. Patient samples Patient samples which had been submitted to SKBL-Tampa for routine chemistry analysis were used for estimates of production capacity.
Propane gas A special grade (99"5%) of propane gas was obtained from Bishop Welding Supply, Tampa, Florida and a 100 pound (net weight) tank was located outside the building to meet fire-code and safety requirements.
Deionized water Laboratory grade deionized water was used throughout and was supplied by the in-house reverse osmosis/deionized water system. Chemistry reagents. All reagents were supplied by Olympus and manufactured by DMA (Data Medical Associates, Inc., 2016 East Randol Mill Road, Arlington, Texas 76011, USA).

Methods
Guidelines for clinical laboratory instrumentation evaluations followed those specified by National Committee on Clinical Laboratory Standards (NCCLS) [1][2][3].
Chemistry methods All methods employ bichromatic measurements. The chemical basis for each method summarized below is taken from Olympus AU5000 application sheets [4]: (1) Na/K: Flame photometry; lithium (3 mmol/1) is used as an internal standard and flame colour is measured at 589 nm (Na), 768 nm (K) and 671 nm (Li).

Precision study
The precision studies were conducted using two levels of control material. Within-run precision was determined from a set of 30 data points. Mean, standard deviation and coefficient of variation were calculated for each test.
Day-to-day precision was determined for two 10-day sets and for the 20-day period overall.
Linearity study The linearity of each method was assessed using either weighed-in standards obtained from New England Reagent Laboratory (NERL), or, in the case of the enzymes, cholesterol, triglycerides and bilirubin, pooled patient sera. Multiple points throughout the dynamic range were measured.

Carry-over experiment
In this experiment, 10 separate aliquots of the high and low controls were sampled to determine a random mean value for each control. This was compared to the carry-over mean value obtained for each. The latter was determined by alternating the sampling of the controls. A carry-over percentage was calculated as follows: Carry-over (%) Random mean Carry-over mean 100 Random mean Correlation study Patient specimens (N 30) were included in calculating correlation coefficients using the method of least squares.
Values obtained on the SMAC II were compared to those obtained on the Olympus AU5061.   Throughput experiment The instrument was first primed with fresh reagents. Calibration samples and controls were placed on the instrument for initial set up and the instrument was re-calibrated every hour. Quality control samples were run every 300 patients. The instrument was operated continuously for a period of 5"3 hours during which 1410 patients were analysed. Four other throughput experiments consisted of from 500 to 600 specimens each. Results

Within-run precision
Data on within-run precision is summarized in tables 2 and 3. Precision for both levels of control material was 1-2% or less for most methods. Bicarbonate (4%) was a notable exception. At low concentrations, creatinine (4"2%) and bilirubin (3"8%) were somewhat higher than the other methods. The enzyme measurements showed remarkably good precision.
Day-to-@ precision Data on day-to-day precision is summarized in tables 4 and 5. Data was accumulated from 10 runs performed on 20 separate days. Group means were calculated from the daily means for each of two 10-day periods and overall for
All the other methods had less than 1% carry-over and for some methods, i.e. glucose, ALT and BUN no evidence of carry-over was found.

Correlation
For most methods, patient values agreed with those of the SMAC II. Correlation coefficients above 0"900 were obtained, except for chloride, bicarbonate and calcium.

Discussion
The instrument is designed for high sample and test throughput. The sample rate of the AU 5061 is fixed. This means that profiles offrom to 26 tests are performed at the same rate. The rate claimed by Olympus is 300/hour. The effective throughput taking into account start-up routines is actually 200/hour for the first hour and a maximum of 280/hour once the instrument is in full operation. Patient throughput was found to be 265 samples/hour. Other instrumental approaches consider test throughput rather than sample throughput. With the Olympus, a conservative throughput of 265 patient samples/hour and a maximum of 26 tests/sample gives a test throughput of 6890 tests/hour. This high throughput makes some special requirements on the steps devoted to specimen processing and data entry both prior to and following analysis. It is imperative that data entry is expedited and the flow of samples to the instrument optimized. Interface to a host computer facilitates the transfer of data.
The number of tests available on the AU 5061 is 26 and includes the two analytes on the flame (Na/K). This leaves 24 compared to 18 (20-Na/K) on the SMAC II.
The six available tests can be utilized effectively by offloading procedures done either manually or on smaller pieces of automated equipment. Additional savings in personnel and laboratory supplies can be achieved in this way. The software has a feature that provides objective photometric readings for assessing the degree of lipemia, icterus and hemolysis of the sample.
The instrument is fast. The throughput is two-and-a-half to three times that of the SMAC II analyser. The limiting factor is processing and essential data entry steps required for getting specimens onto the instrument. The linear range is wider in some cases which means fewer repeats. The 700 mg/dl upper limit for triglycerides, for instance, means that fewer repeats are required. Fewer repeats adds to the cost savings that can be realized. Temperature control for analytical measurements is achieved with a dry bath which surrounds the cuvette wheel. Coolant is circulated through an enclosed system and gives a constant fixed reaction temperature of 37 + 0"2 C. Variation of the reaction temperature to 25 C or 30 C for enzyme measurements is not possible. The glass reaction cuvette also serves as the measurement cuvette.
Up to eight photometric measurements are taken at two wavelengths as the cuvettes advance. A series of fibre  (see table 9). Delayed readings can be specified for optimizing reaction rates for both end-point and kinetic methods. The test parameters include minimum and maximum absorbance values, reagent blanks, quality control, linear range and reference range. Sample blanks can be run but this requires devoting one of the available 24 tests to each blank method. Thus 12 blanks can be run on the AU 5061, together with 12 tests for a total of 24 available channels.
The amount of reagent required for each determination is in the range of 250-500 tl which dramatically reduces the cost of reagents compared to that required to operate continuous-flow instruments. Table 9 summarizes method parameters for the various procedures. In addition, the cost of other consumables (pump tubing, coils etc.) is eliminated. Expenses for AU5061 consumables other than reagents include pump tubes to operate the flame, sample cups, and bar code labels. Other items would include reagent tubing, sample probes, and reaction cuvettes. During the brief 60-day evaluation period, none of the latter items needed replacement.
Approximately 80 ft of floor space are required for installation and operation. Electrical requirements include 220 V (+ 10%), 50/60 Hz (+1 Hz), single phase grounded outlet. Other physical requirements include a floor drain, a source of deionized water capable of delivering 60 l/h, and a supply of propane gas to operate the flame photometer. Deionized water is required for blanking and washing the cuvettes. The water is de-gassed within the system to prevent air bubbles developing in any of the lines. The pumps required for this purpose create a certain amount of noise when they are in operation. However, they are shielded and the noise level was not considered a serious concern.
Test requisitions can be created on the system for up to 4000 patient specimens. This can be performed either before, during (STATs) or after the analyser is opera- M :0"7-1"3 mg/dl F :0"6-1 "2 mg/dl Reagent 1/Reagent 2 when more than one reagent is used. EP Endpoint; Rate Kinetic method, usually with 3-6 readings. Primary/secondary wavelength for bichromatic analysis. Reagents are stored in refrigerated compartments above the analytical components. The bottle size supplied was considered too small for high volume use. The container size should be increased four fold to accommodate 2 for most chemistries. Checks on the existing reagent volumes are a part of the start-up protocol and this information is available through the CRT.
Maintenance is relatively simple to perform. The items requiring attention include the pump tubes on the flame. These should be replaced once a week. Weekly and monthly maintenance requires 30-40 min in each case.
Every three months the conveyor belt and water tank need to be cleaned and this takes approximately 2 h to complete. The evaluation revealed that additional maintenance was required for the calcium, chloride and bicarbonate methods. This requires approximately 20 min/day and involves a cleaning of the reagent lines. The sample probe can become clogged with fibrin clots and must be visually inspected during operation. The instrument performs a forced flush with deionized water between samples which helps to minimize this occurrence. Additionally, time must be devoted to preparing the samples which includes filtering all samples. Replacement of the sample probe is not difficult. The sample probe is a vital link. Since each module contains its own probe, a malfunction in one probe effects only the four tests on that module.
The computer system offers only rudimentary qualitycontrol features. This function is better handled using more sophisticated programs available on a laboratory host computer. The options available include calculations for standard deviation, mean, and coefficient of variation. It will not allow elimination of outliers, and editing of QC data is not easily performed. It does provide Qc charts for each test and each control individually, but decision-making trees and summary reports are not included.
The instrument for evaluation was not interfaced to the laboratory computer. Testing of the bi-directional interface using a PC indicated that it did perform as specified by Olympus. The conclusion reached was that an interfaced program could be written and interfaced to the SKBL host computer system.
The bar-code reading device was not functional with the labels supplied. A source of bar-code labels was obtained and the labels were placed on the rack instead of the tubes. This allows for the re-use of the label and was considered a better approach than placing it on the tube. The instrument requires entry of specimen identification before analysis is performed. The Olympus defaults to not doing any tests but only for that particular sample. As a result, the instrument does not stop completely if a patient identification has not been entered. It will continue to analyse the other specimens that have been entered.
Olympus has indicated that the current software programs are to be enhanced with special packages for quality control, result verification using a multi-variate alogorithm and management reporting. The calculated parameters available at this time are A (B), (A-B)/A, A/B and A/(B-A). It does not permit, for instance, the calculation of an ionized calcium. The bar code reading device worked with the code-a-bar (European) format. It uses an LED device and has the advantage that it does not have any moving parts. Olympus has indicated further that a dial-up capability for diagnostics will be offered.
Training is easily accomplished. Operator proficiency can be achieved in a shorter period than that required to operate a SMAC II. The instrument is menu driven by the CRT and is easy to follow. Help screens are not a part of the software program, however. Trouble shooting is easier to perform using a much improved and updated operator's manual.

Conclusion
The instrument did not have any serious problems during the 60-day evaluation period that caused it to be down. Overall, the Olympus AU5061 was judged to be an effective replacement for the SMAC II instrument providing advantages in increased throughput with significant reductions in operating expenses.