ACR site accreditation for MRI should be out by year's end

March 16, 1994

Public pressure mounts to control modalityAn MRI site accreditation program under development by the AmericanCollege of Radiology could be in place as early as the end ofthis year, according to Dr. William G. Bradley Jr., chairman ofthe ACR's

Public pressure mounts to control modality

An MRI site accreditation program under development by the AmericanCollege of Radiology could be in place as early as the end ofthis year, according to Dr. William G. Bradley Jr., chairman ofthe ACR's committee on standards and accreditation in neuroradiologyand MRI. The ACR is working on accreditation for both MRI andultrasound, modeled after the professional society's successfulprogram for mammography (SCAN 11/17/93).

The ACR's effort to police itself on MRI quality issues mayhelp fend off outside efforts to control technology and price.Some insurers have already sought to discriminate among MRI providerson the basis of quality, focusing on the convenient issue of fieldstrength (see related story below).

Although no policy has been successfully implemented, suchschemes will continue to surface as payers seek the greatest valuefor their outlays. As the competition for subscribers intensifiesamong insurers, quality should also emerge as a marketing tactic.

Increasing public knowledge and consensus about what constitutesquality medical care will drive medical, government and insurerefforts to implement mandated practice standards.

MRI, under attack for its perceived role in fostering health-careinflation as well as medical fraud and abuse, will be a primecandidate for mandated standards. Quality assurance has becomea crucial issue for MRI providers "Clinton has clearly targetedMRI and CT," Bradley said. "The ACR is telling us togo fast on (the MR site accreditation program), because the governmentis going to get involved if we don't."

The ACR MR site accreditation subcommittee that Bradley headswas formed in December 1992, and is composed of radiologists andphysicists who are all members of both the ACR and the Societyof Magnetic Resonance.

Weeding out bad MRI scanners. The stated goal of the ACR'saccreditation program is to identify MRI systems that fail toprovide image quality above a threshold standard, tentativelyset to encompass 85% of systems in current operation. By design,the program is directed only at the specific issue of technicalimage quality, not professional aspects of the diagnostic process.

"We're focusing only on inadequate machines," Bradleysaid, "and not on the competence of interpretation."

The proposed MRI accreditation program, as outlined by Bradleyin the Journal of Magnetic Resonance Imaging, will consist ofthree parts:** an extensive application questionnaire;** an analysisof images of a standard phantom; and** an analysis of clinicalimages.

The application questionnaire will address qualifications ofpersonnel, quality assurance practices, safety and other issuescovered in the ACR Standard of Practice for MRI.

Phantom images will provide measurements of such image characteristicsas signal-to-noise ratio, homogeneity, artifacts and section profile.

Finally, the clinical images section will evaluate studiesof anatomy most commonly imaged: brain, cervical spine, lumbarspine and knee.

Assessment criteria for clinical images will likely includemeasures of coverage, edge definition, contrast and artifacts.

Mid-field MRI passes muster in studyInvestigators at UniversityHospital in London, Ontario, have compared the diagnostic accuracyof MRI systems at 0.5 tesla versus 1.5 tesla. Preliminary resultsindicate that state-of-the-art mid-field systems impose no inherentdiagnostic constraints compared to their more expensive high-fieldcounterparts.

The study provides ammunition for mid-field advocates battlingpayors who would discriminate in their reimbursement policiesagainst lower field strength systems based on quality concerns.

Funded jointly by the University Hospital administration, GEMedical Systems, and MR contrast supplier Berlex Laboratories,the ongoing study includes data from 845 patients, recruited forsix disease categories.

The disease categories included:** multiple sclerosis (221patients);** temporal lobe epilepsy (169 patients);** bronchogeniccarcinoma (19 patients);** pancreatic carcinoma (88 patients);**cervical diskovertebral disease (118 patients); and** internalderangement of the knee (230 patients).

Each patient underwent scanning at both field strengths withina 24-hour period. Three readers blindly interpreted the exams:a neuroradiologist and a body MRI specialist who read all of thestudies, and a radiologist considered expert for the particulardisease category.

A reader-operator-characteristic (ROC) analysis of the datacollected to date for all categories has demonstrated no significantdifferences in accuracy between the mid-field and high-field studiesfor both the experts and the more generalist reader.

"Even at extremely small patient numbers, the (ROC) curveswere basically parallel and almost superimposed on each other,and they have remained the same with some minor fluctuations forthe rest of the recruitment," said Dr. Alexander D. Vellet.

Formerly at the University Hospital, Vellet is now medicaldirector of Western Canada MR Center, Canada's first private MRcenter, located in Calgary, Alberta.

"There is no downside to the quality of health care youcan deliver at 0.5 tesla," said Vellet, who, like most Canadianradiologists, works with a high-field system. "Ironically,although the government had no role in this study at all, theyundoubtedly will benefit from the results, maybe more than anyoneelse."