Location dependent biases in automatic 96-well microplate readers.

Procedures performed in 96-well microplates and quantitated by automatic readers assume instruments to be precise, accurate, and free of well location dependent bias. Instrument specifications generally focus on precision and accuracy without specifically addressing biases which are dependent on well location. These biases appear to be meniscus dependent and can be demonstrated in varying degrees in automatic readers of many designs by using a reverse plate wet test, which compares repeated readings of a dye loaded plate in normal and reversed positions. This test analyzes differences between readings and is, therefore, independent of pipetting error or other experimental variables such as protein binding or immunoassay variability. Different plates increased or decreased the magnitude of observed errors but did not themselves cause the errors measured by the reverse plate wet test. Error patterns were consistent for each reader and varied widely among the 16 instruments tested. Only 4 of 16 instruments passed an existing manufacturer's specification for precision, and only one of the 16 readers tested passed a similar specification for accuracy. The severest location dependent bias was found in an instrument which exhibited excellent repeatability and consistently passed its built-in diagnostic tests. One reader with significant bias was returned to the manufacturer for routine service and calibration, but it was not demonstrably improved. The reverse plate wet test is an extremely useful diagnostic tool for quality control at all stages of instrument manufacture and use.