[Evaluation of pipetting systems. II. Precision and accuracy of precision dispensers]

To evaluate the precision and accuracy of the liquid dispensers being used in our institution.

A total of 15 dispensers (8 micro and 7 macro) of automatic, semiautomatic or manual types were evaluated. They had been in use from 0.5 to 7 years.

The volume dispensed was established by gravimetry using an analytical balance (Chio JP-160) and deionized distilled water as previously described. The microdispensers were evaluated using volumes of 20 and 100 microL, and the macrodispensers with 1 and 5 mL. Each dispenser was evaluated using 10 replicates: the mean and the CV (coefficient of variation) were calculated with the 10 replicates. The mean was transformed to per cent of the theoretical volume. Thus a 100% ratio of accuracy corresponds to a perfect accuracy. An important methodological aspect was that the 10 dispensing replicates were performed personally by the habitual user of the dispenser.

The mean and CV are shown in table 2 for the microdispensers, and in table 3 for the macrodispenser. In the total 25 evaluation in tables 2 and 3 there were 8 instances in which precision was poor (CV above 3%) and also 8 cases of sizable inaccuracy (ratio outside of 100 +/- 5%). In 4 (2 micro and 2 macro) of these 8 inaccuracies, the error ranged from 10% to 22%. A minority was within the specifications of precision and accuracy claimed by the manufacturers of the dispensers with the exception of the manual microdispensers which were within specifications in most instances. We attribute the latter to the lack of a mechanical device in the manual dispensers.

As in a previous evaluation of automatic micropipettes, we detected a high proportion of pipetting systems out of specifications in both precision and accuracy. The magnitude of error in the accuracy of 4 dispensers (10 to 22%) would be catastrophic in methods demanding good pipetting.