When hemolysis changes interpretation: decision-level shifts and reportability failures in routine coagulation...

Hemolysis is common in coagulation testing, yet generic acceptance or rejection rules may not reflect analyte-, decision-level, and reportability-specific effects. We assessed graded hemolysis across clinical decision strata on the Sysmex CS-5100. Residual citrated plasma was pooled into five decision strata for PT/INR, aPTT, Clauss fibrinogen, and D-dimer (≥20 specimens/pool), spiked with standardized hemolysate to final hemolysis levels of 0-5% (0, 0.5, 1, 2, 3, 4, 5%), and measured in five replicates. Hemolysis severity was characterized using CS-5100 sample-check HI/flags, XN-1000 cell-free hemoglobin, and Roche cobas c702 hemolysis index. Percent bias versus the matched non-hemolyzed baseline was evaluated against EFLM-derived desirable bias specifications and RCV-based allowable change limits. Hemolysis effects were analyte- and decision-level dependent. PT/INR showed overall limited effects, although selected confidence intervals crossed desirable bias limits; all PT results remained within RCV-based limits. aPTT showed a biphasic, stratum-dependent pattern, with prolongation in lower strata and shortening in prolonged strata. Fibrinogen showed positive bias, but progressive loss of reportability, with earlier 'No coagulation' in low-fibrinogen strata, was more clinically relevant. D-dimer showed modest, consistent negative bias, mainly exceeding desirable bias in the lowest stratum; no above-cutoff pool shifted below the 0.50 mg/L FEU deep vein thrombosis cutoff. Across all assays, reportable results remained within RCV-based limits. Hemolysis management should integrate analyte- and decision-level effects, numeric reportability, instrument-derived HI/flags, and clinical context rather than rely on a universal rejection rule.