Effect of hs Troponin I Measurement Uncertainty on Interpretation at Clinical Decision Limits
Abstract
Aim: The aim of this study was to calculate the uncertainty of the hs-cTn I test in which the clinical decision limit value was important and to evaluate the possible effect of the calculated measurement uncertainty for the test result interpretation at the clinical decision limits.
Material and Methods: Measurement uncertainty was calculated according to Nordtest technical report 537. The clinical decision limits of hs-cTnI were 2 ng/L, 6 ng/L, 19 ng/L and 100 ng/L. Lower and upper limit values of hs-cTnI test results reported by our laboratory between May 1 - October 31 2018 were calculated according to uncertainty value. When the hs-cTn I test results are reported together with the lower and upper limit values according to the uncertainty value, the number and percentage of test results that can be interpreted differently in the clinical decision limits are determined.
Results: The uncertainty of within-laboratory reproducibility (uRW) value of hs-cTn I test was 4.89% and the uncertainty component from the certified or nominal value (uCref) was 3.88%. The measurement uncertainty (95% confidence interval) of hs-cTn I was estimated to be 14.02%. The number of hs-cTn I test results analysed for suspected myocardial infarction was 2299. When the lower and upper limit values of hs-cTn I results are added to the laboratory result report according to the calculated uncertainty value, the total number of results that can be interpreted differently in the clinical decision limits is 358 (15.6%). As a result of providing uncertainty value information in the laboratory results report, we determined that 7.35% of the results reported below the decision limits can be interpreted above the decision limit and 8.22% of the results reported above the decision limits can be interpreted as below the decision limits.
Conclusion: Reporting uncertainty of measurement affects the clinical decision process. We believe that our study, in which we evaluate possible decision changes for reported results with measurement uncertainty data, supports this view.