RESUMO
OBJECTIVE: To establish a norm of the precision achievable with a micropipette in an IRMA assay under routine conditions. MATERIAL AND METHODS: A micropipette (Gilson) adjusted to dispense 100 microL was used by a single analyst with experience in its use. In each assay, ten aliquots of radioactive antiprolactin were pipetted in clean tubes (PRE-batch tubes), followed by pipetting of the tubes being processed in the assay, and at the end, a second pipetting of 10 aliquots in clean tubes (POST-batch tubes). The study includes the data of 15 consecutive batches during a seven month period with an overall mean of 283 tubes per batch. The PRE- and POST-tubes were read in a gamma counter (Crystal plus). The mean, SD and CV for PRE, POST and global (PRE+POST) tubes were calculated for each batch. RESULTS: The global CV of the 15 batches ranged from 1.6 to 6.9%, mean of 3.1%. We found no evidence of increased imprecision due to fatigue of the analyst, but surprisingly, we observed that in nine of the 15 batches there was a significant difference in the means of the PRE-tubes vs the POST-tubes (t test) without differences in precision. Thus, part of the global variability is due to what we have called pseudoimprecision (i.e. an increase in CV due to differences in means). In addition, the POST-tubes had higher values in the first 7 batches but the opposite occurred in the last 8 batches (table 2). This shift in the sign of the PRE-POST differences suggests the presence of opposite factors operating in time, i.e. one or more factors increased the volume of pipetting after using the pipette more than 150 times (batches 1-7) whereas other/others decreased it (batches 8-15). CONCLUSIONS: 1. Our first approximation to a norm of micropipetting precision in batches of 200-300 tubes was a CV of 3.1%. 2. This norm was influenced by a problem of pseudoimprecision detected ex-post-facto. 3. Our findings justify continuation studies to detect the pseudoimprecision and evaluate its causes prospectively.