Comparative evaluation of a Technicon SMAC2/RA1000 System with an American Monitor Parallel during normal service work

To meet the requirements of an increasing biochemical work-load, manufacturers have developed large and expensive computer-controlled analytical systems. Therefore the choice ofmain analyser for a clinical biochemistry department is a major decision with long-term financial implications. Advice is usually sought from current users, but this is often more anecdotal than factual and certainly not of a comparative nature. For many large departments, the choice is between two systems employing different analyt!cal concepts. These are the Parallel (American Monitor, UK) and a combination of the SMAC2 and RA1000 analysers (Technicon, UK).


Introduction
To meet the requirements of an increasing biochemical work-load, manufacturers have developed large and expensive computer-controlled analytical systems. Therefore the choice ofmain analyser for a clinical biochemistry department is a major decision with long-term financial implications. Advice is usually sought from current users, but this is often more anecdotal than factual and certainly not of a comparative nature. For many large departments, the choice is between two systems employing different analyt!cal concepts. These are the Parallel (American Monitor, UK) and a combination of the SMAC2 and RA1000 analysers (Technicon, UK).
The Parallel is a recently introduced 30-channel, highcapacity, discretionary analyser, whilst the SMAC2 is an established 12-23 channel continuous-flow nondiscretionary analyser which can only be made selective in terms of reporting. The RA1000 is a bench-top discretionary analyser operating like the Parallel but with much less capacity. The SMAC2 and RA1000 together form a system of similar capability to the Parallel. With both a Parallel and a SMAC2/RA1000 combination installed in the authors' separate departments in Leeds, a study has been undertaken to provide comparative data which should aid potential purchasers. On one day a week, over a period of 15 weeks, each laboratory reconstituted one bottle of each material. The contents of the bottles were analysed randomly 10 times throughout the day for the commonly requested tests (see table 1). On each subsequent week the day was advanced by one in order that each of the five working days were covered on three occasions. For tests less frequently assayed, two aliquots of each material were measured during one analytical run per week (see table 2).
For the investigation of reliability, a detailed log of the daily working parameters, electromechanical and other failures were recorded throughout the whole three-month period. The relative running costs ofthe two systems were calculated for the whole of 1984, including staff, consumables and maintenance contracts.

Analytical performance
The values obtained on the quality-control materials were analysed using the Statistics Package for Social Sciences (SPSS). Analysis ofvariance on the data showed that significant imprecision was between, rather than within, analytical runs. For each test, the overall mean, standard deviation and coefficient ofvariation is given in tables and 2. Results are shown only for those tests common to both systems and where a test level was close to the detection limit of the assay, the results have been excluded from the tables. The data were reanalysed after removal of possible outliers using Healy's procedure [1] and the recalculated coefficients of variation are shown in brackets in tables and 2.

Costs
Details of capital costs can be found in a comprehensive review of large analysers based on manufacturers' information [2]. An estimate ofthe 1984 running costs for   LGI samples for the SMAC2/RA1000 are processed by reception staff. These procedural differences have been allowed for by calculating the number ofclerical and reception staff at the LGI employed on direct SMAC2/ RA1000 work on a pro rata basis.

Reliability
Data collected throughout the three-month period on the work-load and reliability are given in table 4. 208

Discussion
The Parallel was designed to operate in a completely selective discretionary mode. However, at St. James's Hospital, because of a reluctance by clinical staff to request tests individually, three main profiles for urea and electrolytes, liver function and bone were offered, whilst remaining tests were measured individually. The order in which the samples were analysed was found to be important because, prior to any test analysis,, appropriate reagent lines were purged if not immediately preceded by another analysis for the same test. This meant that the reagent consumption was much higher than expected and necessitated the grouping of the expensive tests where possible: either within the analytical run or on a less frequent batch basis. In particular, the high cost of the bicarbonate reagent discourages the dispersion of non- * Imprecisions of the analysers not significantly different (P > 0"05).

Precision
During the study period both the SMAC2 and RA1000 performed better than the Parallel and the latter's imprecision did not greatly improve on exclusion of possible outliers. These findings for the SMAC2 are in accordance with data published in the Wellcome Quality Assurance Scheme end-of-term report (October 1985), which shows that the SMAC2 analyser group have lower imprecisions than the Parallel group for corresponding channels. However, the report shows the RA1000 and Parallel groups to be more comparable, which is not reflected in our data. Since the period of the comparison the performance of the St. James's Parallel has improved, reflected by external quality assurance scheme returns (table 5).

Reliability
There were considerable problems with the mechanical reliability of the Parallel during the period of evaluation, the instrument being totally unusable on two full days. The problems were due to a computer failure necessitating a replacement computer being installed overnight by American Monitor, and, secondly, an electromechanical failure in the vial wash system. A further major contribution to the excessive downtime was the occasional halting of the Parallel during analysis due to a communications failure between the two system computers (this has been rectified by American Monitor). The downtime of the SMAC2 was predominantly due to replacing membranes during a run. The RA1000 downtime was approximately 15 min per week and was contributed to by lamp and sample probe problems.

Costings
In this study, during which little discretionary requesting was exercised or encouraged in either hospital, the total cost of running the two systems was very similar. Because of the considerable difference in the way the two hospitals handle similar work-loads, comparison of overall cost of service is more appropriate. Theoretically it might seem to be possible to reduce the overall cost at St. James's Hospital by encouraging a higher degree of selectivity in requests but as previously stated, this would be countered by increased reagent costs.