Week 4 Method Verification (Linearity)

Overview of Method Verification

• Method verification refers to the process of confirming that a laboratory test or method produces reliable results within a specified range.

• The verification process discussed includes assessments of linearity using standards and patient samples.

Types of Linearity Assessments

Part A: Linearity Using Standards

• Purpose: To ensure that the BCP method is linear across the concentration ranges relevant for various analytes.

• Range of Interest:

  • The range depends on the analyte. For example:

  • TSH (Thyroid Stimulating Hormone) and Troponin: Require readings from 0% to 150% of the measuring range.

  • Albumin: No necessity to measure near zero; focus is from 10 grams per liter upwards.

• Standard Concentration Range: 50% to 150% of the reference interval is considered significant for albumin.

• Data Collection: Duplicates of absorbance readings are required for each standard concentration measured.

Part B: Linearity Using Patient Samples

• Objective: To ensure that patient samples behave similarly to aqueous standards, confirming that the assay method’s performance can be generalized beyond standards.

• Importance of Matrix Differences: Serum matrix differs from an aqueous solution, potentially affecting results.

Instructions for Data Analysis

Data Preparation for Part A

• Absorbance Values: The only section where absorbance values are utilized directly.

• Conversion: For remaining sections (Part B and beyond), convert absorbance values to concentrations before analysis.

Data Analysis Steps

1. Plotting: Create a scatter plot of mean absorbance vs. concentration of standards.

   • Determine whether the relationship appears linear.

2. Regression Analysis: Perform regression analysis to calculate the R squared value.

   • Interpretation of R squared: Values above 0.95 indicate good correlation.

   • Example R squared: 0.99 indicates strong linearity.

3. Residuals Plot: Construct a residuals plot to visually confirm linearity.

   • The residuals are the differences between observed values and the predicted values (Y-hat).

   • Residuals plot interpretation:

     • Ideally should show a random pattern around zero.

     • Non-random patterns may indicate non-linearity.

Conclusion of Part A

• Confirm if the assay demonstrates linearity and identify the linear range.

• Compare observed linearity results to manufacturer specifications provided in appendix.

Data Analysis for Part B (Patient Samples)

Sample Preparation

• Create five different pools from two initial patient pools, measuring their albumin concentrations.

• Expected values should be listed based on created pools.

Data Analysis Steps for Part B

1. Mean Calculation: Plot the mean of duplicate samples' albumin concentrations.

2. Assessment: Perform linear regression on the observed versus expected concentrations.

   • Maintain a visual assessment of data linearity.

   • Calculate R squared for the linear regression line.

3. Residuals Plot: Similar to Part A, plot residuals to validate linearity of the data relationship.

4. Percent Difference Calculation:

   • Calculate percent difference using the formula: $ext{Percent Difference} = rac{ ext{Residual}}{Y ext{-hat}} imes 100$

   • Acceptability criteria: Percent difference must be ( \leq 5\% ) based on CLIA standards for total allowable error for albumin.

   • Confirm results for all pools.

Conclusion of Part B

• Confirm linear relationship of patient samples against standards from Part A.

• Confirm all percent differences are within the acceptable range.

General Notes

• Data Interpretation: It's crucial that students correctly interpret their results regardless of their similarity to other colleagues’ data.

• Use of Software: Utilizing Excel for calculations and plotting is encouraged for accuracy and efficiency.

• Reporting Findings: Clearly present findings, especially linearity assessments and comparisons to references in the appendix.

Additional Considerations

• Variability in results is normal; focus should be on accurate data interpretation and presentation. There are no absolute “correct” answers in method verification reports.