Summary
A retrospective study of 10,634 percutaneous coronary interventions found radiation-alert threshold breaches in 0.93% of procedures and 2.1% of complex procedures. The authors developed an internally assessed score using early clinical and procedural variables.
A medRxiv preprint analysing 10,634 consecutive percutaneous coronary intervention (PCI) procedures reports that a composite radiation-alert threshold was exceeded in 99 procedures, or 0.93% of the total. The rate was 2.1% during complex PCI, compared with 0.4% during standard PCI.
The researchers also developed a pragmatic risk score using clinical and procedural information available early in an operation. Its purpose is to identify procedures at higher risk of reaching radiation-alert levels while there is still time to activate dose-sparing strategies.
How the study defined a radiation alert
PCI uses X-ray fluoroscopy to guide catheters through the blood vessels and treat narrowed or blocked coronary arteries. Radiation exposure during a procedure can be described using several measures. The study used a composite endpoint in which any one of three thresholds qualified as radiation overexposure: a dose-area product above 500 Gy·cm², cumulative reference-point air kerma above 5 Gy, or fluoroscopy lasting more than 60 minutes.
Dose-area product combines radiation output with the area exposed. Reference-point air kerma is a dose-related measure reported for a defined location near the patient, while fluoroscopy time records how long continuous or intermittent X-ray guidance was used.
The analysis covered procedures performed from January 2008 through May 2018. Of the total, 3,457 procedures, or 32.5%, were classified as complex. The definition included intervention on an unprotected left main artery, rotational atherectomy, treatment of a chronic total occlusion, or bifurcation or trifurcation PCI.
What the score found
In a multivariable analysis, higher body mass index, male sex, chronic total occlusion treatment, rotational atherectomy, bifurcation intervention and treatment of at least two vessels independently predicted higher reference-point air kerma. These findings identify patient and procedural characteristics associated with greater radiation dose in the study population.
The risk score was built from early clinical and procedural variables rather than information available only after a large dose had accumulated. It showed an area under the receiver-operating-characteristic curve of 0.772, with a 95% bootstrap confidence interval of 0.725 to 0.818. This indicates good separation between procedures that did and did not reach the composite alert endpoint within the analysed dataset.
The score was also reported as well calibrated, with a Brier score of 0.0091 and an observed-to-expected ratio of 0.994. Calibration describes how closely predicted risk corresponds to the frequency of events observed in the data.
The study is a retrospective observational cohort, not a clinical trial or prospective interventional study, and the score was internally assessed using the analysed procedures. Its immediate contribution is a planning-stage method for identifying higher-risk cases. Prospective use would be needed to assess how score-guided dose-sparing decisions affect radiation exposure and patient outcomes. The findings are reported in a medRxiv preprint.