A Quality Improvement Initiative Employing Size-Based Reference Levels for Monitoring CT Dose Appropriateness
Abstract
Purpose
CT reference levels are mandated by Texas state regulations, and Medical Physics is responsible for reviewing CT studies that exceed these levels and determining the cause. The current clinical reference level for body CT, 60mGy, is applied uniformly across all patient sizes, resulting in dose alerts being predominantly triggered in larger patients. To increase patient safety and avoid unintended or accidental exposures, we implemented size-specific Medical Physics Reference Levels (MPRL) and conducted a quality improvement initiative.
Methods
Routine chest and portal venous CT exams were split into seven size groupings by water equivalent diameter (WED) as determined by the radiation dose index management software (RDIM). For each group, the MPRL was defined as the CTDIvol separating outliers from the exponential trendline of the previous year’s patient data acquired across multiple CT scanners. A medical physicist reviewed each study which exceeded the MPRL, selected a predefined category describing the primary reason for the higher than expected dose index, and documented the investigation within the RDIM.
Results
A total of 385 Chest and Abdomen/Pelvis exams exceeding MPRL values were reviewed by nine qualified medical physicists. 44% had arms down and lateral anatomy truncation visible in the localizer. 41% were patients with high WED variability axially, had hardware implants, and/or had lateral anatomy truncation in the localizer. Suboptimal patient positioning accounted for 4% of cases, while incorrect protocol settings were responsible for 3%. The remaining studies were incorrectly flagged by the RDIM or had insufficient data to analyze.
Conclusion
Overall, 85% of exams exceeded the MPRL due to patient habitus or positioning constraints beyond the technologist control and did not imply unintended radiation overexposure. Our results suggest that alternative metrics beyond size-based CTDIvol reference levels may be needed to efficiently identify the approximately 7% of studies associated with potentially correctable issues.