Clinical Decision-Making Suffers from Inequivalent Measurement Results and Inadequate Reference Intervals.

Abstract

Appropriate clinical decision-making relies on the interpretation of equivalent measurement results in the context of valid clinical decision limits. Besides guideline-based decision limits, reference intervals (RIs) are commonly used to discriminate between abnormal results and results from "healthy" individuals. This study evaluated the suitability of RIs in light of the analytical bias for laboratories in the Netherlands using one standardized, one harmonized, and one unharmonized measurand (creatinine, hemoglobin, and ferritin, respectively). Three types of data were collected: (a) external quality assessment (EQA) performance data from the Dutch Foundation for Quality Assurance in Laboratory Medicine (SKML); (b) the RIs reported by laboratories for a 55-year-old female; and (c) harmonized RIs established by using unique routine patient results using RefineR. Routinely used RIs (b) were compared to the harmonized RIs (c) and evaluated in combination with the analytical bias at the lower and upper reference limits. Laboratories reported a variety of routinely used RIs that were inconsistent with the analytical bias, with differences between measurement procedures. The use of assays that perform within allowable bias limits does not automatically guarantee that the appropriate RI is used, allowing potential for structural misinterpretation of important diagnoses in patients. The use of RIs that are inconsistent with the analytical bias causes unnecessary between-laboratory differences in clinical decision-making. Adopting harmonized RIs facilitates similar interpretation of results across facilities. Harmonized RIs can be adopted immediately if the observed bias is acceptable or eliminated, or after standardization/harmonization of measurands without complete metrological traceability.