METHOD VALIDATION - THE LINEARITY OR REPORTABLE RANGE EXPERIMENT
Elsa F. Quam, BS, MT(ASCP)
METHOD VALIDATION - It is important to assess the useful analytical range of a laboratory method, i.e., the lowest and highest test results that are reliable and can be reported. Manufacturers make claims for the reportable range of their methods by the stating the upper and lower limits of the range. It is generally assumed that the analytical method produces a linear response and that the test results between those upper and lower limits are then reportable. CLIA recommends that laboratories verify the reportable range of all moderate and high complexity tests [1].
CLIA does not strictly require that a method provide a linear response, but laboratories commonly use a "linearity" type of experiment to check the reportable range for a test. In this experiment, a laboratory will analyze a series of samples with known concentrations, or a series of diluted samples. The measured results are plotted on the y-axis vs the expected or known values on the x-axis. The reportable range is then assessed by drawing the best straight line through the linear portion of the data. That line may be drawn manually by making a "visual fit", or by computer using the "best fit" from linear regression statistics. The statistical analysis could become fairly complicated to assess linearity, but it is commonly accepted that the reportable range can be adequately determined from "eyeballing" the data and manually drawing the best straight line that fits the lowest points in the series.
The National Committee for Clinical Laboratory Standards (NCCLS) recommends a minimum of at least 4 - and preferably 5 - different levels or concentrations [2]. More than 5 can be used, but we have found that 5 levels are convenient and almost always sufficient.
Standard solutions may be easy to prepare for some tests. For others, manufacturers and proficiency testing agencies may supply linearity sets having known values or known relationships between materials. Dilutions of patient specimens, or pools of patient specmens, may also be used and are often convenient and economical when high values are available. In some cases, e.g., TDMs, it may be necessary to spike a pool with the analyte to be measured in order to get the desired high level.
Selection of the diluent is important because the matrix of the specimen needs to be maintained, e.g., when serum is the usual specimen type, it may be important to maintain the serum concentration in the whole series of test samples. For general chemistry tests, water and saline will often be okay. For other tests, it may be better to use bovine or serum albumin preparations, specimens with low concentrations, or drug free serum. One way to decide on diluent is to follow the manufacturer's recommendation of the diluent to use with out-of-range specimens.
It is convenient to use two pools - one near the zero level or close to the detection limit and the other near or slightly above the expected upper limit of the working range. Determine the total volume needed for analysis, select appropriate volumetric pipets and then do the following:
If more levels are desired, this dilution protocol can be changed, e.g., the two pools could be mixed 4 to 1, 3 to 2, 2 to 3, 1 to 4 to give four intermediate levels for a total of six levels for the experiment.
NCCLS recommends making 4 measurements on each specimen or pool. In practice, we have found that 3 replicates are generally sufficient. This also means triplicates on the original high and low pools.
Plot the mean of the measured values on the y-axis versus the known value on the x-axis. Visually inspect the plot for a linear relationship. Manually draw a straight line through as many of the points as possible and estimate the reportable range.
Worksheet 1 for Validation of Reportable Range: See this worksheet for directions on preparing the samples, recording the results, and preparing a simple plot for visual assessment of reportable range. This worksheet can be used as a general worksheet for any test.
Cholesterol example Worksheet 1: See this example for a practical application of worksheet 1 for a cholesterol test.
Worksheet 2 for Quantifying Errors: See this worksheet for directions on additional calculations for quantifying errors.
Cholesterol example Worksheet 2: See this example for a practical application of worksheet 2 for a cholesterol test.
