P060] The effect of technical parameters in practice on abdominal and computed tomography (CT) dose

Abstract

Purpose Radiation doses for computed tomography (CT) examinations are highly variable. Understanding how technical parameters impact variation in practice can guide future optimization strategies. Methods As part of a PCORI and NIH funded study, we created a CT radiation dose registry on 1.85 million CT adult (>=15 years) exams prospectively collected between January 2015 and June 2017 from 151 institutions across seven countries. We estimated the relative contributions of technical parameters to mean dose length product (DLP) and proportions of high-dose examinations for abdomen CT, after adjusting for patient, institutional characteristics, and scanner model, using hierarchical and logistic regression, respectively. Results Mean DLP and proportion of high-dose examinations varied modestly (10–50%) by institutional characteristics and scanner model after adjusting for patient characteristics (age, sex and size). In multivariable models, dose variation was primarily attributable to institutional uses of technical parameters. When accounting for the technical parameters, the differences between institutions were attenuated or eliminated. In our data, mAs counted for the largest difference in dose across country, and was twice as important as the other factors. For each standard deviation change in the log-transformed mAs, the average DLP increased by 60%. Variation in the use of multiphase scanning was the next most important factor explaining variation in DLP. With each standard deviation change in the log-transformed phase count, the average DLP increased by 51%. A typical change in mAs and phase count had twice the effect on dose as typical changes in other parameters. For every increase in log transformed standard deviation change in technical parameters, there was a 23% increase in DLP for scan length, a 21% decrease for pitch, an 18% increase for kvp, and a 4% decrease for slice thickness. In order of magnitude, the technical parameters most influential on abdominal dose were: mAs, phase count, scan length, pitch, kvp, and slice thickness. Conclusions Differences between abdomen doses across institutions were largely attenuated when controlling for dose parameters, to a degree not seen when controlling for patient, institutional characteristics, and scanner model.