Electricity Grid Carbon Intensity-Not Imaging Utilization-as the Primary Source of Variation in Greenhouse Gas Emissions for MRI and CT.

Abstract

<b>Background.</b> MRI- and CT-associated greenhouse gas (GHG) emissions are determined not only by electricity consumption but also the carbon intensity of the power grid supplying such electricity, in turn reflecting local fuel mixtures. <b>Purpose.</b> To assess the relative contributions of imaging utilization and grid carbon intensity to state variation in per-beneficiary GHG emissions attributable to MRI and CT in Medicare Part B fee-for-service beneficiaries. <b>Methods.</b> This retrospective study used national and state-level data from calendar-year 2022 federal datasets. Aggregate MRI and CT service counts in Medicare Part B fee-for-service beneficiaries were obtained from the CMS Medicare Physician and Other Practitioners by Geography and Service file. Medicare Part B FFS enrollment was determined using the CMS Medicare Monthly Enrollment file. Cardon dioxide-equivalent (CO₂e) emission rates, measuring GHG emissions associated with electricity generation as an indicator of grid carbon intensity, were obtained from U.S. Environmental Protection Agency data. Per-examination electricity consumption for MRI and CT was obtained from a prior study that extracted scanner data using specialized equipment. GHG emissions were estimated as the product of imaging-associated electricity use (i.e., the product of aggregate service counts and per-examination electricity consumption) and grid carbon intensity. Relative contributions of utilization-weighted electricity use and grid carbon intensity toward variation in per-beneficiary GHG emissions were assessed by log-linear variance decomposition. <b>Results.</b> MRI and CT exhibited national utilization per 1000 beneficiaries of 271 examinations and 805 examinations, respectively; associated estimated electricity use of 157,734 MWh and 28,236 MWh, respectively; and associated estimated GHG emissions of 54,052 MT CO₂e and 9991 MT CO₂e, respectively Estimated GHG emissions per 1000 beneficiaries nationally were 2.19 MT CO₂e, ranging from 0.05 MT CO₂e (Vermont) to 4.82 MT CO₂e (Missouri). In log-linear decomposition, grid carbon intensity accounted for 76.3% of state variation in per-beneficiary GHG emissions, utilization-weighted electricity use for 16.3%, and their covariance for 7.4%. <b>Conclusion.</b> Per-beneficiary GHG emissions varied widely across states, primarily due to variation in grid carbon intensity rather than utilization-weighted electricity use. <b>Clinical Impact.</b> The findings highlight the role of grid decarbonization and energy procurement strategies in complementing imaging stewardship for reducing imaging's environmental footprint.