Normalization of activated partial thromboplastin time and diluted Russell’s viper venom time ratios can be safely omitted in the assessment of lupus anticoagulant

TL;DR

This study shows that a step in testing for lupus anticoagulant can be safely skipped without affecting diagnosis accuracy.

Contribution

The study demonstrates that normalization of LA ratios can be omitted without compromising diagnostic performance.

Findings

Non-normalized LA ratios showed high discriminative performance with area under the curve values of 0.9818 and 0.8926.

Optimal cut-off values of 1.304 and 1.554 for dilute Russell’s viper venom time and activated partial thromboplastin time were identified.

Applying non-normalized cut-offs resulted in no reclassifications in the validation dataset.

Abstract

Normalization of lupus anticoagulant (LA) ratios is recommended by both the Clinical & Laboratory Standards Institute and the International Society of Thrombosis and Hemostasis to reduce inter-assay and intra-assay variation. This study evaluates if the normalization step can be safely omitted.

This retrospective observational study evaluated the diagnostic performance of non-normalized LA ratios. Anonymized patient data from routine clinical workflows were divided into 2 datasets: one for establishing optimal non-normalized cut-off values (n = 440) and one for validation (n = 306). Clinical classification (positive/negative) in the first dataset served as the reference standard, enabling the establishment of optimal cut-off values for the non-normalized LA ratios using receiver operating characteristic calculations and Youden’s J statistic with bootstrapping. These optimal non-normalized cutoffs were then used on the validation set to compare the number of reclassifications when compared with normalized LA ratios.

Receiver operating characteristic analysis demonstrated high discriminative performance of non-normalized LA ratios for both dilute Russell’s viper venom time (area under the curve 0.9818; 95% CI, 0.9462-1) and activated partial thromboplastin time (area under the curve 0.8926; 95% CI, 0.8077-0.9775). Youden’s J identified optimal cut-off values of 1.304 for dilute Russell’s viper venom time (95% CI, 1.297-1.307) and of 1.554 for activated partial thromboplastin time (95% CI, 1.504-1.562). Applying the above criteria to the validation dataset resulted in no reclassifications.

The normalization step in the analysis of LA ratios can be safely omitted, allowing for a reduced laboratory workload, without affecting result interpretation and thus clinical outcomes.

•Laboratories adjust lupus anticoagulant tests to reduce variation.•We studied large amounts of hospital data to test if this adjustment can be simplified.•The new approach gave results identical to the current standard.•Simplifying the process does not affect diagnosis and might further lessen the variation.

Laboratories adjust lupus anticoagulant tests to reduce variation.

We studied large amounts of hospital data to test if this adjustment can be simplified.

The new approach gave results identical to the current standard.

Simplifying the process does not affect diagnosis and might further lessen the variation.

Figures