Determining Analytical Performance Characteristics

Introduction

Occasionally we get asked, "What is the error at 1 ppb?" or "How sure are you of this result?". The answers are highly dependent on the sample in addition to the analytical technique. Unless we have performed method validation on your sample or similar samples, we can only estimate the answer. Validation of analytical methods is the measurement of performance characteristics such as

Specificity 

How can we determine if the sample contains positive or negative interferences? Quantitative recovery of sample spikes is a good indication that the method works well for the sample. However we can not afford to spike every sample type with every analyte, and often must rely on surrogate spikes to give us an indication. The presence of interferences can often be determined by comparing mass spectra or UV spectra across the profile of a chromatographic peak, i.e. determining 

Recent articles from WCAS discussed the lack of specificity for aromatic hydrocarbons by EPA Method 8020 when gasoline, mineral spirits, or even crude oil was present. While this is an example of the lack of specificity experiments being documented in the method, it is also an example of application of a method to problems which we believe the method was not designed to handle.

Robustness 

Methods provided by many agencies, especially USP, NIOSH, and EPA, are written in recipe fashion with no discussion of the chemical principles or allowable variations. In our opinion a good method should describe the analytical principles as well as the results of intentional variation of testing parameters such as sample storage, sample preparation, and chromatographic conditions. This is a tremendous aid to the analyst, specifying limits to experimental conditions and allowing some latitude when modifications are needed.

Detection Limits 

Labs wishing to validate methods should determine Method Detection Limits (MDLs). MDLs are supposed to be the lowest concentration at which we can be 99% confident that the compound is in a sample. But MDLs for EPA are determined in clean water, and this has little to do with the detection limit in a real sample. In our opinion MDLs simply demonstrate that an analyst can perform the method in a suitable manner. Sample detection limits should be determined not only from the MDLs, but other factors as 

well, such as sample dilution factors, interferences, and recovery of sample spikes. For more information on detection limits....

These are just a few of the items that need to be considered in validating methods.   Let us prepare a more detailed proposal for you.   If you have questions concerning the accuracy or precision of your results, call Dr. Northington to discuss validation of test methods with your samples. Performance characteristics are generally sample dependent. 

For a quotation...