An approach to selecting auto-verification limits and validating their error detection performance independently for pre-analytical and analytical errors

Abstract

BackgroundAuto-verification limits are widely used to trigger confirmatory actions to enable detection of pre-analytical, analytical and post-analytical errors. An approach is presented for validating auto-verification limit performance in a laboratory-specific manner, independently for pre-analytical and analytical error detection. MethodsTo evaluate this approach, MA Generator (www.huvaros.com) was used to run error-detection simulations using various upper–limit checks (ULC) and lower–limit checks (LLC). Pre-analytical error detection was defined as triggering of a limit check alarm within one erroneous result. Analytical error detection was defined as triggering a limit check alarm within the scheduled internal QC measurement interval, both with ≥97.5% probability. Furthermore, the limit check alarm rates were obtained. ResultsPre-analytical error detection and rapid detection of larger analytical errors by limit checks outperformed moving average quality control at the cost of a significantly larger number of alarms. A pre-analytical error detection by LLC and ULC of ≥−55% and >60%, ≥−10% and ≥20%, and ≥−40% and ≥50% and an analytical error detection of ≥−4% and ≥15%, ≥−3% and ≥4% and ≥−30% and ≥25% were obtained for hemoglobin, sodium and calcium, respectively. ConclusionsThe obtained ULC and LLC alarm rate and error detection performance, enabled substantiated selection of optimal auto-verification limits and validation thereof.