Prediction of precision for purity methods

Abstract Methods that determine the relative purity of biopharmaceuticals represent the most widely used form of analysis for the pharmaceutical industry. The ability to rapidly assess method capability, or the uncertainty of measurements, under actual use conditions continues to present significant challenges. We have refined and applied the model of Uncertainty Based on Current Information (UBCI) to predict the precision of the purity measurements and compared the predicted precision to the statistically derived variability for several different types of purity methods. The determination of the statistically derived method variability was derived from the analysis of data sets ranging from hundreds to thousands of measurements for each different method type. The predicted precision was found to be in excellent agreement with the statistically obtained values with R2=0.94. This demonstration of concurrence between the predicted and actual precision provides an opportunity for streamlining laborious conventional (statically derived) assessment of method precision and leveraging the UBCI model utilizing much smaller data sets or even a single experiment.