Documenting the Budget

The uncertainty budget, ideally, should contain all of the components below. The presentation of the budget is down to personal preference of the scientist performing the assessment, be it on an assay by assay basis or with all results for each step displayed together.

Statement of Measurement Uncertainty

This contains the details of:

  • The measurand definition
  • The contributing uncertainty influences
  • The mathematical model applied (if applicable)
  • The associated calculations
    • Type A variation
    • Type B evaluations
  • The associated combinations
  • The final result

Budget contents

  • The person performing the assessment
  • The date performed
  • Expected review date
  • All results

For uncertainty contributors the reports should include the following in addition:

  • Sensitivity coefficients
  • Probability distribution
  • Uncertainty Index (see below)

Number of Significant Digits

The measurement uncertainty reported for a measurand should be consistent to the number reported in final results. In some scenarios the number of significant figures in the final result is driven by the MU determination but in order to maintain pathology laboratory results as clinically interpretable this is not implemented. A maximum of two significant figures is suggested by the GUM document and applied in most cases.

In cases where rounding of uncertainty results is required to achieve the desired number of significant figures, uncertainty values should always be rounded up to ensure underestimation of uncertainty is not encountered.

The uncertainty index and the contribution of each component

Once the budget has been constructed a simple calculation of the influence each individual component has on the overall uncertainty is possible. This is useful as if the review of the budget suggests that it is unacceptably high, the contributor with the highest proportion of uncertainty in the overall budget would be the component that it would be indicated to resolve first.

Move on to Suitability and Targets

%d bloggers like this: