Occupational radiation protection in interventional radiology: A multicenter multimethod study of anesthesiology personnel

Occupational radiation protection in interventional radiology depends not only on individual knowledge and training but also on protective behavior, dosimetry, equipment availability, and institutional support. We evaluated these domains among anesthesiology personnel using a multimethod design that combined a formally developed, expert content-validated, and pilot-tested questionnaire with center audit and structured procedure observation. This four-center cross-sectional census study included 82 eligible non-pilot personnel. Objective knowledge and self-reported protective practice were assessed by the instrument. Structural availability was assessed by center audit, and 50 interventional-radiology procedures were directly observed; 20% were double-observed. A prespecified psychometric branching rule determined whether practice could be represented by a composite score. The mean Objective Knowledge Score was 6.16 ± 1.58 of 10. The nine practice items did not support a single composite construct: corrected item-total correlations were all below 0.30 and ordinal alpha was 0.40. Self-reported adherence was 49.4% for lead glasses, 43.8% for protective-shield use, and 45.5% for personal-dosimeter use. Neither formal training nor Knowledge Score was significantly associated with these three prespecified outcomes after Holm correction. Center audits showed variable structural support, while direct observation demonstrated generally acceptable inter-rater agreement except for lead-glasses use (κ = 0.55). Descriptive triangulation identified center- and construct-specific discordance between knowledge, self-report, observed behavior, and audited availability. Radiation-protection practice in interventional radiology was therefore not adequately represented by a single adherence score; combining behavioral and structural measurement sources provided information that questionnaire data alone would not capture.