EP16 refers to “Measurement Procedure Comparison and Bias Estimation Using Patient Samples.” It’s a guideline provided by the Clinical and Laboratory Standards Institute (CLSI) that focuses on the comparison of measurement procedures and the estimation of bias using patient samples in clinical laboratories.
The EP16 guideline outlines the process of comparing the results obtained from two or more measurement procedures (analysers or methods) using real patient samples. The goal is to assess the agreement or bias between these procedures and to identify any significant differences that could affect patient care or clinical decision-making.
Key aspects covered in EP16 include:
- Study Design: The guideline provides recommendations for selecting appropriate patient samples, specifying the number of replicates, and considering factors such as patient population characteristics and clinical conditions.
- Data Analysis: EP16 explains how to perform statistical analyses to determine the bias and identify any systematic differences between measurement procedures. It covers methods such as the paired t-test, Deming regression, Bland-Altman analysis, and others
- Calculation of Bias: The guideline provides guidance on calculating bias, which indicates the degree of agreement or difference between measurement procedures. Bias is a crucial parameter to consider when comparing methods.
- Estimation of Total Allowable Error: EP16 discusses the concept of total allowable error (TEa) and how to interpret the calculated bias in relation to TEa to determine if the differences between methods are clinically significant.
- Data Interpretation: The guideline helps laboratory professionals interpret the results of the comparison study and make informed decisions about the clinical impact of differences between measurement procedures.
- Reporting: EP16 outlines the necessary information that should be included in the final report of the comparison study, ensuring transparency and traceability of the process
Overall, EP16 is a valuable resource for clinical laboratories aiming to evaluate and compare measurement procedures to ensure accurate and reliable patient test results. It helps laboratories make informed decisions about method selection, method validation, and continuous quality improvement efforts.
Here’s a hypothetical example of the verification process outlined in EP16, using data from a comparison study between two different measurement procedures (Method A and Method B) for a specific analyte:
Analyte: Glucose
Patient Samples: 20 patient samples with varying glucose concentrations
Measurement Procedures:
- Method A (Current Method): Analyzer X
- Method B (New Method): Analyzer Y
Here’s a simplified version of the data and the steps involved in the verification process:
- Sample Collection and Preparation: Collect 20 patient samples covering a range of glucose concentrations. Ensure that the samples are anonymized and blinded for the study.
- Analysis of Patient Samples: Run each patient sample on both Method A and Method B. Record the glucose concentration results obtained from each analyser.
| Sample | Glucose (Method A) | Glucose (Method B) |
| 85 mg/dL | 83 mg/dL | |
| 112 mg/dL | 110 mg/dL | |
| 150 mg/dL | 152 mg/dL | |
| 200 mg/dL | 198 mg/dL | |
| …… | ||
| 20 | 270 mg/dL | 268 mg/dL |
- Data Analysis: Perform statistical analyses to assess the agreement between Method A and Method B.
- Calculate Bias:
Calculate the bias for each sample: Bias = Glucose (Method A) – Glucose (Method B).
- Calculate Mean Bias:
Calculate the mean bias across all samples.
- Calculate Standard Deviation (SD) of Bias:
Calculate the standard deviation of bias.
- Deming Regression Analysis: Perform Deming regression analysis to determine the relationship between Method A and Method B and assess any systematic differences between the methods
- Bland-Altman Plot: Create a Bland-Altman plot to visualize the agreement between Method A and Method B. The plot should show the differences in glucose concentrations between the two methods on the y-axis and the mean glucose concentrations on the x-axis
- Interpretation: Based on the bias, Deming regression, and Bland-Altman plot, interpret the agreement between Method A and Method B. Assess whether the observed differences are clinically significant and within acceptable limits defined by the laboratory’s allowable total error (TEa).
- Reporting: Prepare a detailed report summarizing the study design, data collection, analysis methods, results, and conclusions. Include any necessary recommendations for action based on the findings.
Please note that this is a simplified hypothetical example. In a real-world scenario, the data analysis would involve more rigorous statistical calculations and considerations, and the interpretation would depend on the specific clinical context and laboratory requirements.
About the author
Dr. Sambhu Chakraborty is a distinguished consultant in quality accreditation for laboratories and hospitals. With a leadership portfolio that includes directorial roles in two laboratory organizations and a consulting firm, as well as chairmanship in a prominent laboratory organization, Dr. Chakraborty is a respected voice in the field. For further engagement or inquiries, Dr. Chakraborty can be contacted through email at director@iaqmconsultants.com and info@sambhuchakraborty.com. Additional resourcesand contact information are available on his websites, https://www.quality-pathshala.com and https://www.sambhuchakraborty.com, or via WhatsApp at +919830051583