Capturing the Flag – Leveraging Laboratory Measurements of Uncertainty to Accelerate Data Usability
Oral Presentation
Prepared by H. Leibovitz1, J. Krueger1, R. Runyon1, S. Kolb2, K. Miller2, T. Smith3
1 - CSC, 15000 Conference Center Drive, Chantilly, VA, 20151, United States
2 - CSC, 6361 Walker Lane, Suite 300, Alexandria, VA, 22310, United States
3 - U.S. EPA, 1200 Pennsylvania Ave, NW, 5104A, Ariel Rios Building / Room B452, Washington, DC, 20460, United States
Contact Information: [email protected]; (401) 742-8922
ABSTRACT
The data review community uses an à la carte approach to choose quality control indicators that support validation/verification of measurement quality objectives. Often these quality control indicators are quality control data measurements associated with a specific analytical batch of field samples/quality control samples. Examples of specific indicators include data that represents precision, accuracy, sensitivity, selectivity, and representativeness. Laboratory measurement uncertainty is a unique quality indicator that reflects statistically a smorgasbord of multiple measurements reflecting multiple quality indicators over time. Reporting uncertainty not only provides a confidence level necessary in the decisions based upon particular guidance levels, but also a unique measure of a laboratory’s capability over time. Radiological results typically include uncertainty and reporting uncertainty is necessary to comply with ISO 17025. However, due to a number of issues, often the reporting of uncertainty with environmental results is inconsistent and lacks a standardized approach. A standardized approach must identify best practices and define which components of laboratory uncertainty should be included, which measurements best represent which component, the number of measurements included in a calculation, the desired level of confidence for the uncertainty, and finally the statistical calculation itself. This paper borrows and expands upon an approach used by the Naval Sea System Command to automate the calculation and reporting of uncertainty without exhaustive “re-tooling” and collection of additional data. An Excel template automates the calculations of uncertainty using data from replicate traditional quality control indicators and input of variables. Furthermore, a “nested” approach provides the data consumer the option to specify which time collected quality control measurement indicators to evaluate in order to represent the contribution of different components of the uncertainty of laboratory measurements. Examples include laboratory method uncertainty represented by the laboratory control sample and laboratory project matrix uncertainty represented by matrix spike/duplicates data.
Oral Presentation
Prepared by H. Leibovitz1, J. Krueger1, R. Runyon1, S. Kolb2, K. Miller2, T. Smith3
1 - CSC, 15000 Conference Center Drive, Chantilly, VA, 20151, United States
2 - CSC, 6361 Walker Lane, Suite 300, Alexandria, VA, 22310, United States
3 - U.S. EPA, 1200 Pennsylvania Ave, NW, 5104A, Ariel Rios Building / Room B452, Washington, DC, 20460, United States
Contact Information: [email protected]; (401) 742-8922
ABSTRACT
The data review community uses an à la carte approach to choose quality control indicators that support validation/verification of measurement quality objectives. Often these quality control indicators are quality control data measurements associated with a specific analytical batch of field samples/quality control samples. Examples of specific indicators include data that represents precision, accuracy, sensitivity, selectivity, and representativeness. Laboratory measurement uncertainty is a unique quality indicator that reflects statistically a smorgasbord of multiple measurements reflecting multiple quality indicators over time. Reporting uncertainty not only provides a confidence level necessary in the decisions based upon particular guidance levels, but also a unique measure of a laboratory’s capability over time. Radiological results typically include uncertainty and reporting uncertainty is necessary to comply with ISO 17025. However, due to a number of issues, often the reporting of uncertainty with environmental results is inconsistent and lacks a standardized approach. A standardized approach must identify best practices and define which components of laboratory uncertainty should be included, which measurements best represent which component, the number of measurements included in a calculation, the desired level of confidence for the uncertainty, and finally the statistical calculation itself. This paper borrows and expands upon an approach used by the Naval Sea System Command to automate the calculation and reporting of uncertainty without exhaustive “re-tooling” and collection of additional data. An Excel template automates the calculations of uncertainty using data from replicate traditional quality control indicators and input of variables. Furthermore, a “nested” approach provides the data consumer the option to specify which time collected quality control measurement indicators to evaluate in order to represent the contribution of different components of the uncertainty of laboratory measurements. Examples include laboratory method uncertainty represented by the laboratory control sample and laboratory project matrix uncertainty represented by matrix spike/duplicates data.
