Prediction intervals on Deming regression are a major new feature in the Analyse-it Method Validation Edition version 4.90, just released.
A prediction interval is an interval that has a given probability of including a future observation(s). They are very useful in method validation for testing the commutability of reference materials or processed samples with patient samples. Two CLSI protocols, EP14-A3: Evaluation of Commutability of Processed Samples and EP30-A: Characterization and Qualification of Commutable Reference Materials for Laboratory Medicine both use prediction intervals.
We will illustrate this new feature using an example from CLSI EP14-A3:
1) Open the workbook EP14-A3.xlsx.
2) On the Analyse-it ribbon tab, in the Statistical Analysis group, click Method Comparison and then click Ordinary Deming regression.
The analysis task pane opens.
3) In the X (Reference / Comparative) drop-down list, select Cholesterol: A.
4) In the Y (Test / New) drop-down list, select Cholesterol: A.
5) On the Analyse-it ribbon tab, in the Method Comparison group, click Restrict to Group.
6) In the Group / Color / Symbol drop-down list, select Sample Type.
7) In the Restrict fit to group drop-down list, select Patient.
8) In the Prediction band edit box, type 95%.
NOTE: Select the Familywise coverage check box to control the probability of simultaneously for all additional samples rather than individually for each sample.
9) On the Descriptives task pane, select Label points, and choose Additional groups only.
10) Click Calculate.
The report shows the scatter plot with fitted regression line and 95% prediction interval (see image below). The regression line is only fitted to the points in the Patient group, as set in step 7 above, and additional points are colored depending on the type of sample, as set in step 6 above.
Any points outside the prediction band are not commutable with the patient samples, and in this case you can see sample ‘c’ is not commutable. The commutability table shows the additional samples and whether they are commutable or not with the patient samples.
The steps to perform an EP30 study are the same as described above. You should note that EP30 forms the prediction interval using the fit of the patient samples and the precision of the reference materials, where-as Analyse-it uses the fit and precision of the patient samples. We chose to implement it like this since there are usually too few reference material samples to establish a reliable estimate of the precision.
We have extended the prediction intervals beyond the CLSI EP guidelines, so they support any number of replicates and are also available with Ordinary and Weighted Deming regression. This alleviates the need to log transform values as is recommended in EP14, which, although it corrects the constant CV, distorts the relationship between the two methods.
If you have active maintenance you can download and install the update now, see updating the software. If maintenance on your license has expired you can renew it to get this update and forthcoming updates, see renew maintenance.
Let us know your comments, opinions and any questions you have on this post. Your feedback is always appreciated.