Reducing patient-specific QA workload through statistical process control and complexity metrics

Patient-specific quality assurance (PSQA) for VMAT treatments represents a significant workload. This study presents an integrated approach combining plan complexity metrics with statistical process control (SPC) to safely reduce PSQA burden while maintaining robust process monitoring.

We analyzed 557 consecutive patients (695 VMAT plans, 1650 arcs). MCSv and SAS were combined into a composite modulation index (CMI). Spearman correlation analysis identified the optimal gamma criterion among 22 configurations, with Bonferroni correction across 220 pairwise tests. Statistical distribution analysis determined tolerance limits using the percentile-equivalent method. ROC analysis established a complexity threshold for PSQA exemption, validated on an independent cohort of 267 patients.

High correlation was observed between (1 - MCSv) × SAS(7.5 mm) and 3%/1.5 mm local gamma criterion (r_S = -0.74, p < 10^-11). The beta distribution provided superior fit to PSQA data across all standard goodness-of-fit metrics (KS = 0.155 vs 0.229-0.235; AD = 34.3 vs 147-160) compared to normal, lognormal, and gamma distributions. Control and action limits were established at 78.1% and 66.5%, and ROC analysis demonstrated excellent discriminative performance (AUC = 0.90, 95% CI: 0.87-0.92). A complexity threshold of 0.131 achieved 100% sensitivity (95% CI: 93.4%-100.0%). Independent validation confirmed perfect sensitivity with zero false negatives. Eight months of clinical implementation achieved 26% workload reduction with no safety incidents.

This methodology demonstrates that integrating complexity metrics with appropriate statistical modeling and SPC enables safe, clinically validated PSQA workload reduction while maintaining rigorous quality standards and continuous process monitoring.