Establishing denominator-consistent Six Sigma benchmarks for venous blood specimen rejection: a multi-center retrospective quality-indicator study using 21 million specimens in Western China.

BACKGROUND

Pre-analytical venous blood specimen rejection can delay clinical decision-making and may increase laboratory workload, yet multicenter benchmarking remains limited by inconsistent rejection definitions and denominator selection. This study aimed to support denominator-consistent benchmarking, transferable implementation, and quality monitoring across heterogeneous-resource settings.

METHODS

We conducted a retrospective quality-indicator study across 10 hospitals in Western China using routine venous blood specimen data collected from January 1, 2023, to April 30, 2024. Rejection reasons were standardized with a prespecified mapping dictionary. Center-level rejection rates, defects per million opportunities (DPMO), and Six Sigma values were calculated using monthly venous specimen volume as the primary denominator. Percentile-based benchmarks, pooled estimates, and a month-adjusted random-intercept generalized linear mixed-effects model were used to assess performance and residual between-center heterogeneity.

RESULTS

Among 21,319,200 venous specimens, 42,060 were rejected, yielding a pooled rejection rate of 0.197%. The leading standardized rejection categories were clot/aggregation, insufficient or incorrect volume, and container/tube error. Rejection patterns varied across centers and clinical settings: inpatient and emergency events were dominated by clotting and volume problems, whereas outpatient and physical examination events more often showed order, timing, or identification errors. After temporal adjustment, measurable between-center heterogeneity persisted (intraclass correlation coefficient = 0.045).

CONCLUSION

This multi-center quality-indicator study provides a reproducible, denominator-consistent framework for benchmarking venous blood specimen rejection across multiple centers. Percentile-based benchmarks and standardized reason spectra may support local implementation and transferable Six Sigma monitoring in healthcare systems with heterogeneous resources.