Estimating nurses' exposures to ionizing radiation: the elusive gold standard.

This study assessed ionizing radiation exposure in 58,125 registered nurses in British Columbia, Canada, for a cohort study of cancer morbidity and mortality. Two methods were used: (1) a survey of nurses in more than 100 acute care hospitals and health care centers; (2) and monitoring data reported to the National Dose Registry of Health Canada, considered the gold standard. The mean exposure of cohort nurses monitored during the study period from 1974 to 2000 was 0.27 milliSieverts (7028 person-years of monitoring). Of 609,809 person-years in the cohort, 554,595 (90.9%) were identified as unexposed by both exposure assessment methods. Despite crude agreement of 91% between the methods, weighted kappa for agreement beyond chance was only 0.045, and the sensitivity of the survey method to capture National Dose Registry monitored person-years was only 0.085 (specificity = 0.97). The survey missed exposures outside the acute care setting. The National Dose Registry also missed potential exposures, especially among hospital emergency department and pediatric staff nurses. It was unlikely that either method estimated nurses' true exposures to ionizing radiation with good sensitivity and specificity. The difficulty in exposure assessment likely arises because fewer than 10% of registered nurses are exposed to ionizing radiation, yet the settings in which they are exposed vary tremendously. This means that careful hazard assessment is required to ensure that monitoring is complete where exposures are probable, without incurring the excess costs and lack of specificity of including the unexposed.