Why is the Alara software needed for calculation of the measure?
As a radiology measure, the measure derives standardized data elements from structured fields within both the electronic health record (EHR) and the radiology electronic clinical data systems, including the Radiology Information System (RIS) and the Picture Archiving and Communication System (PACS). Primary imaging data including Radiation Dose Structured Reports and image pixel data are stored in the PACS in Digital Imaging and Communications in Medicine (DICOM) format, a universally adopted standard for medical imaging information. Because of limitations in their specifications and format, eCQMs cannot currently access and consume elements from these radiology sources in their original DICOM formats. Thus, translation software was developed to transform primary data into a format that the eCQM can consume. This eCQM requires the use of additional software (translation software) to access the primary data elements that are required for measure computation and translate them into data elements that can be ingested by this eCQM. The purpose of this translation software is to access and link these primary data elements with minimal site burden, assess each CT exam for eligibility based on initial population criteria, and generate the three data elements mapped to a clinical terminology for eCQM consumption: CT Dose and Image Quality Category, Calculated CT Size-Adjusted Dose, and Calculated CT Global Noise.
The data inputs have been defined in LOINC, specified in measure submission materials to the MUC and NQF, and published on the eCQI Resource Center. The framework for classifying CT scans into CT categories has been published, An Image Quality-informed Framework for CT Characterization, Radiology, 2022. Details regarding the measure were developed with close involvement of the Technical Expert Panel, are specified in the NQF application packet (Patient Safety fall 2021 meeting) including the definition of all variables, and their mapping to LOINC codes. Definitions for inclusion and exclusion, how missing data and sample size considerations were handled, as well as values to consider exams out of range based on size adjusted dose and noise have been provided. These details were reviewed by the NQF Scientific Methods Panel, and by the Patient Safety Committee. The measure was judged to have very high scores for measure reliability, validity, feasibility, and useability. Some of the details highlighted in the application and discussed through addressed during the NQF evaluation.
What is the cost of the software?
While the Alara Imaging Software 8 for CMS Measure Compliance is proprietary, it will be available to all reporting entities free of charge and accessible by creating a secure account through the measure steward’s website. Despite no cost to obtain Alara software, there are resource costs to integrating it into various technologies. Hospital staff will need to oversee installation, configuration, and ongoing management of software.
Although the Alara software will be free to providers, what is the cost and burden associated with implementing and managing things going forward? Including costs associated with hardware, application support, and software maintenance.
The pre-processing software has been designed specifically to decrease health system implementation costs. The software accepts a wide range of FHIR, HL7 formats for EHR data and DICOM CT radiation dose and image data to eliminate burden. The measure has been developed identically to other eCQMs using proven formats: QDM (for immediate implementation) and FHIR (when adopted in the future) implementation in accordance with CMS’s aim of encouraging interoperability based on FHIR APIs. Thus, the overall burden is comparable to existing eCQMs.
To clarify the reporting process, we note that a hospital can log in through the measure developer’s secure portal and run the Alara Imaging Software for CMS Measure Compliance inside the firewall. The software runs automatically to create the three intermediate data elements needed for the measure: CT Dose and Image Quality Category, Calculated CT Size-Adjusted Dose, and Calculated CT Global Noise. Once the software finishes creating these intermediate variables, hospitals can send the data to its EHR for measure calculation and reporting. The software allows additional options such as the ability to send the data to other business associates of the hospital if needed. No manual data entry is required.
We anticipate that some EHR vendors may develop solutions to ingest these calculated variables and calculate the eCQM, as they have done for other eCQMs. This burden to EHR developers should be similar to any other new eCQM adopted into the CMS quality reporting programs. We additionally note that the adoption timeline and option to self-select reporting on this measure should provide sufficient flexibility for those hospitals that may need time to integrate the software and implement this measure.
Translation software will be provided free of charge to support improved security, calculation consistency, and reduced burden for reporting entities. During measure testing, the UCSF measure development team interviewed staff (site PIs, PACS administrators, and IT and radiology-IT staff) from diverse field-testing sites and found low burden for measure calculation, similar to the burden for other eCQMs, and generally less than the effort involved in participating in national registries. The implementation imposed no burden on clinicians nor impacted clinical workflows. All data elements are generated during the ordinary course of care. The pre-processing software has been designed specifically to decrease health system implementation costs. Any health system or vendor can receive either tabular translation variable exports or QRDA I / QRDA III file exports to support a reporting entity with measure submission to CMS.
With regards to health equity, why isn’t the measure adjusted for race, ethnicity, or socioeconomic factors?
To the extent they have been studied, social factors including sex, race/ethnicity, and socioeconomic status are not predictive of radiation dose for CT exams. However, patients living in poverty are at higher risk for chronic, comorbid conditions associated with exposure to multiple scans over time and increased cumulative exposure to ionizing radiation from diagnostic imaging. Thus, it is particularly important to ensure that the doses used for CT in these individuals are not excessive, because vulnerable patients are at greatest risk of chronic disease and more likely to be exposed to many irradiating exams.
A common concern is that older CT machines cannot achieve radiation doses as low as more modern equipment. Does this disadvantage imaging facilities with older equipment?
Using the UCSF International CT Dose Registry, which contains over 8 million CT exams from more than 160 imaging facilities, the measure developer has determined the important contributors to radiation dose and its variation. (Smith-Bindman, BMJ, 2019) This is the best information currently available to understand drivers of dose. The results show that patient size, the reason for the CT scan, and the machine (make and model) do influence doses, but do not drive the variation in dose. The dose variation remains after accounting for these factors. The most important determinant of dose is not the machine, but the local decisions by radiologists, medical physicists, and radiology technologists about how to perform the exam and the choice of technical parameters. Appropriately dosed scans can be obtained on all patients and on all machine types.
The decision to use multiple phase exams when a single phase would suffice is the most important contributor to excessive doses. Old machines are capable of performing single-phase exams.
This measure is necessary to ensure that all patients receive optimal, not excessive, radiation doses – no matter where they are imaged or on what machine.
Will hospitals and radiologists know how to use the aggregated measure score to improve quality?
In an NIH-funded randomized controlled trial involving 100 imaging facilities, the developer demonstrated that providing audit feedback to imaging facilities about their doses can achieve meaningful and sustained dose reduction. (Smith-Bindman, JAMA Intern Med, 2020) These measures are intended to provide actionable feedback similar to what was given in the trial. This feedback is stratified by separate CT categories so that providers can pin-point particular areas where they have greater than expected out-of-range exams and can target the areas they need most to improve.
Since there is a relationship between radiation dose and image quality, incentivizing lower radiation doses can lead to unacceptable image quality.
Image quality is not currently a problem with CT imaging; most CT scans (>99%) have acceptable image quality based on testing data assembled in almost 50,000 CT exams as part of measure testing. In order to ensure that image quality is not compromised in the future, a measure of image quality, noise, is included as a balancing measure. The measure of noise is meant to establish a minimum floor for image quality. The developer and steward will monitor the image quality closely and take corrective action if a quality issue arises. It is the physician’s responsibility to ensure that CT image quality is adequate for diagnosis and their responsibility to repeat scans of poor quality. The role of the physician to ensure image quality is acceptable will not change through the use of the measure.
There are multiple ways to measure patient size; why did the measure developers select their approach?
Indeed, there are multiple ways to measure patient size. The primary reason for size adjustment in this measure was to ensure that patient size is not driving failure rates, and to ensure that providers who care for larger patients are not penalized for doing so. The adjustment removes size as a factor for facility out-of-range rates. The NQF Scientific Methods Panel was satisfied with the quality of the adjustment approach.