Multicentre evaluation of the Boehringer Mannheim/Hitachi 747 analysis system

Analytical performance and practicability of the new Boehringer Mannheim/Hitachi 747 analysis system were assessed in a multicentre evaluation involving four laboratories. The analytical performance was evaluated according to a protocol similar to the ECCLS guidelines and comprised 13 analytes including enzymes, substrates and electrolytes. About 65,000 results were obtained within three months. The evaluation was planned and supported by a program system called "Computer Aided Evaluation". Acceptance criteria have been established for judging the results. The median of the within-run coefficients of variation (CVs) in control sera of all methods was below 1%, being far below the acceptance limit of 2%. The median of CVs of between-days imprecision was below 2% (acceptance criterion 3%). The high degree of precision prompted us to set up a biometrical model suitable for the differentiation between deviant points, outliers and measurements that can still be explained by the system performance. No relevant drift effects were observed during eight hours. The methods were linear over a wide range, avoiding rerun analysis in most cases. No sample-related carry-over was found. Reagent-dependent carry-over outside the acceptance limits was measured from uric acid to phosphorus to a slight extent, and from triacylglycerols to lipase, as well as from total protein to bilirubin to a perceptible degree. It can be avoided by separating these reagent combinations in the channel arrangement. Taking a systematic deviation of more than 10% as unacceptable, four of the 13 analytes suffered from interference by haemoglobin, one by bilirubin and one by turbidity. The Boehringer Mannheim/Hitachi 747 analysis system is capable of determining serum indices which in combination with the interferogram allow an assessment of the interference. With the exception of chloride the recovery of the assigned values for all control sera showed values between 95 and 105%. Out of 40 method comparison studies for enzymes and substrates, 31 yielded regression equations with less than 5% proportional errors and less than 5% constant errors. Deviations exceeding these acceptance criteria can be explained by differences in the reagent formulation, in the method employed or in calibration. The agreement of the ISE method comparisons was within a +/- 5% deviation over a wide analytical range. Practicability of the Boehringer Mannheim/Hitachi 747 analysis system was assessed with the help of a questionnaire, in which properties of the instrument were quantified, thus permitting a relatively objective rating. The 190 questions were placed in 14 groups, each dealing with an attribute of the instrument.(ABSTRACT TRUNCATED AT 400 WORDS)