Effect of automation on prothrombin time test in NEQAS surveys.

The performance of coagulometers in the National Quality Assessment Scheme (NEQAS) surveys of the prothrombin time conducted between 1986 and 1987 was reviewed. There were sufficient data for analysis for eight types of coagulometer used with a single type of thromboplastin reagent and one instrument with an alternative reagent. The overall reliability efficacy of each instrument was evaluated by determining the orthogonal regression slope parameters for prothrombin time (PT) and international normalised ratios (INR) using the manual technique as the reference method. Seven of the eight types of coagulometer tended to overestimate the INR. A pattern frequently observed with coagulometers, and difficult to regulate, was a trend to underestimate INR below 3.0 and overestimate higher INR. Overestimation of INR values over 3.0 was particularly pronounced with three types of instrument (Fibrintimer, Lancer, KC4/10). The KC4/10 was used by a sufficient number of participants to permit analysis of the performance of individual instruments. Within instrument differences were similar to those produced by different types of coagulometers. Thromboplastin reagents affected the INR values obtained with coagulometers. The study indicates that each local reagent-instrument combination must be calibrated by the participant to obtain reliable INR values. The use of a general correction factor for a local PT system seems to be invalid owing to the considerable variation in performance of individual coagulometers. The two best guides to the choice of coagulometer may be the deviation from the manual result and precision estimated by the coefficient of variation of the INR.