Different approaches for estimating measurement uncertainty: An effective tool for improving interpretation of results

Background: The estimation of measurement uncertainty (MU) in clinical laboratories is of crucial importance in improving laboratory testing quality and correctly interpreting results. However, it is difficult for clinical laboratories to reliably estimate MU since current guidelines and standards fail to clearly define and harmonize methods to be used for this purpose. Aims: To propose a model for MU estimation in relation to test interpretation on the basis of three different scenarios. Methods: MU estimation was evaluated regard to the inclusion of imprecision and bias components for different test purposes. Results: Three scenarios were identified. The expanded uncertainty values were: 0.56 ng/L and 1.86 ng/L for troponin I at levels 4.95 ng/L and 20.60 ng/L, respectively; 9.4 mu mol/L and 27.2 mu mol/L for creatinine at levels 121.0 mu mol/L. and 390.0 mu mol/L, respectively. 0.2 mmol/L and 0.3 mmol/L for potassium at levels 4.03 mmol/L and 6.35 mmol/L, respectively; 0.36 mmol/L for glucose at level of around 6 mmol/L. These values represent MU results estimated for the three scenarios, which contemplate test results used for 1a) short term patient monitoring, 1b) longer term patient monitoring, 2) comparison with reference intervals and 3) comparison with a clinical decision point. Conclusions: Goal-based measurement uncertainty estimation in clinical laboratories provides an opportunity to improve laboratory testing quality and reduce risk in the interpretation of clinical results.