How To QC MRI Safety Standards

Accreditation is required to assure safety

In the United States, either enterprise or modality accreditation is required for eligibility for reimbursement for services provided to Medicare, Medicaid, or Tricare federal beneficiaries. As a result, it is estimated that over 90% of all MRI providing hospitals and imaging centers in the US hold accreditations that cover their MRI services. Healthcare accreditation standards are meant to be assurances of minimum levels of each quality and safety. In this discussion we’re going to totally ignore the quality of an MRI exam, image, or treatment, and focus exclusively on the promise of MRI safety.

Safety: at a minimum, prevent injury

There are lots of ways that one can measure safety, but perhaps the most elemental is the prevention of injury. Programs that promise safety should, at a minimum, prevent preventable injury.

In the US, serious injury is the minimum mandatory reporting trigger for adverse events involving an FDA-regulated medical device, including MRI (link to FDA MedWatch webpage). While there are reports within the US FDA’s MedWatch database of ‘near misses’ and non-injury accidents, the overwhelming majority of adverse event reports that appear the US FDA database and are classified under the MRI product code (“LNH”) are reports of injury to patient, caregiver, visitor, or MRI service personnel. (link to FDA adverse event search page)

MRI injury trends

In the past two decades, if we tracked the growth in MRI procedure volume in the US, it would show remarkably consistent steady, moderate increases from the year 2000 to the present (with the notable exception of the recent COVID years). If we similarly tracked the change in MRI-classified adverse event reports to the US FDA, if we were reducing the number of adverse events (i.e., injuries), we would expect to see the growth curve of the adverse events be below the procedure volume curve. If we were doing no better -but no worse- in MRI safety, we would expect the change in adverse events to essentially lay immediately on top of, or ‘overdraw,’ the line representing the growth in MRI procedures. Sadly, neither of those are what the data actually shows…

No, what the data actually shows is that the rate of MRI-classified adverse events over the past two decades has been growing at a rate much greater than the rate of procedure volume growth. In fact, if you do a linear regression of both datasets and compare the slopes of the two regressions, the accident rate growth is ~2.8 times steeper than the MRI exam volume growth. Yes, this does vary depending on which year you choose as your starting point, but using each year between 1996 and 2004 as the starting point the slope differentials range from a low of 1.14x (1997) and a high of 5.82x (2004), with an average of 2.98x (indicating that using the year 2000 as the starting point is roughly average, and a reasonable origin point for this longitudinal evaluation.

One proffered explanation for the growth in reported MRI adverse events is that the incidences of MRI accidents and injuries aren’t increasing, but the growth in the numbers is the product of slow-and-steady improvements in adverse event reporting rates. The whipsaw drop-and-rebound in MRI procedure volumes from COVID give us an excellent natural experiment with which to test the ‘better reporting’ theory. If each year sees a slightly better reporting rate than the previous, if there was a precipitous drop in MRI imaging volumes (35% in 2020), we would expect to see a smaller percentage drop in reported MRI adverse events, which wasn’t the case (38% in 2020). But adverse events are a much smaller number, and therefore more subject to more significant annual percentage shifts, so the 2020 data may be less about refuting the ‘better reporting’ theory and more about annual variability. 2021, however, should be more persuasive in refuting the ‘better reporting’ theory, because it saw a 30% annual increase in the number of MRI exams completed but a 13% decrease in the number of reported MRI adverse events.

(More on the rate comparison methodology available here.)

While decades of persistent growth of reported MRI accidents and injuries is something of an indictment of accreditation’s promise of safety by itself, in order to identify the degree to which accreditation might do better we need to have a better understanding of how MRI injuries are occurring and whether or not there are effective strategies for preventing these injury accidents, or are they just an inescapable cost to MRI services?

How MRI patients get injured (and how to stop those injuries)

In 2012 I presented a paper (link to recording) at the RSNA annual meeting (link to abstract) that sought to answer some of these questions. The paper, co-authored with Dr. Emanuel Kanal, specifically evaluated MRI injury accidents that had been reported to the US FDA to determine if existing best practices (taken, at the time, from the 2007 ACR Guidance Document on MR Safe Practices) could have successfully prevented the described injury accidents.

To do this, Dr. Kanal and evaluated two years of MRI-classified adverse event reports from the US FDA (calendar years 2009 & 2010). The top three categories of MRI-specific injury (we discounted falls, fingers pinched in table movement, and other non-MRI-specific injuries) were RF burns, magnetic projectiles, and auditory injury. For each injury type we evaluated three best practices that could prevent that injury type:

Burns (70% of evaluated injury accidents in the 2012 paper)

• • Air gap / padding between patient and transmitting RF element
  • Removal of electrically conductive materials (use of MR Conditional materials, if indicated)
  • Padding to prevent large-caliber body loops

Projectiles (12% of evaluated injury accidents in the 2012 paper)

• • ACR 4-zone facility layout
  • Use of ferromagnetic detection screening
  • Conspicuous safety labeling of objects within MRI suite

Hearing Damage (11% of evaluated injury accidents in the 2012 paper)

• • Require the use of hearing protection
  • Verify hearing protection fit-and-function prior to exam (not part of 2007 ACR Guidance Document, not considered in injury reduction calculation)
  • Provide alternative means of hearing protection (not part of 2007 ACR Guidance Document, not considered in injury reduction calculation)

In that 2012 paper we found that 97% of the burn injuries, 94% of the magnetic projectile injuries, and 29% of the hearing damage incidences would have been prevented had only the existing best practices from the 2007 ACR Guidance Document been used (we did not count the two hearing damage practices that we proposed adding in the mitigation totals). Of the 154 MRI injury accidents assessed, the seven evaluated best practices would likely have prevented 129 (84%) of the injury accidents.

In 2018 I repeated the methodology of the prior study for Metrasens (link to results infographic), using a more contemporary two years of US FDA data (calendar 2015 & 2016), producing similar results.

Burns (55% of evaluated injury accidents in the 2018 paper)

• • Air gap / padding between patient and transmitting RF element
  • Removal of electrically conductive materials (use of MR Conditional materials, if indicated)
  • Padding to prevent large-caliber body loops

Projectiles (23% of evaluated injury accidents in the 2018 paper)

• • ACR 4-zone facility layout
  • Use of ferromagnetic detection screening
  • Conspicuous safety labeling of objects within MRI suite

Hearing Damage (16% of evaluated injury accidents in the 2018 paper)

• • Require the use of hearing protection
  • Verify hearing protection fit-and-function prior to exam (not part of 2007 ACR Guidance Document, not considered in injury reduction calculation)
  • Provide alternative means of hearing protection (not part of 2007 ACR Guidance Document, not considered in injury reduction calculation)

In that 2018 paper I found that 94% of the burn injuries, 69% of the magnetic projectile injuries, and 11% of the hearing damage incidences would have been prevented had only those same existing best practices from the 2007 ACR Guidance Document been used (again, I did not count the two hearing damage practices that we proposed adding in the mitigation totals). Of the 106 MRI injury accidents assessed, the seven evaluated best practices would likely have prevented 78 (74%) of the injury accidents.

(In this two years of injury accident data all 31% of the projectile injury accidents that would not have beep prevented by the identified best practices were injuries to service personnel. In other words, the three identified preventions would have been 100% effective in all clinical care situations.)

Broadly, the injury accident type proportions closely agreed with those published in the 2019 “MRI-related FDA adverse event reports: A 10-yr review“, though the authors of that paper did not exclude the non-MRI-specific “mechanical injuries” from their tally. The FDA paper also did not evaluate best practices’ ability to prevent the injury accidents.

Accreditation and the prevention of MRI injury

Whether an 84% injury accident prevention rate (2012) or a 74% injury accident prevention rate (2018), or an average of the two (79%), it is clear that a small number of existing best practices could make a substantially-positive effect on MRI injury rates and totals. But if these injury-reducing best practices are widely accessible and from standard-setting documents, surely they are requisite elements of enterprise-level and modality-level accreditation in the US, right?

Wrong.

Burns

When it comes to the most frequent source of patient injury, burns, none of the three best practice protections that very nearly eliminate burns (94% & 97% in the two studies) are minimum requirements of any the major accrediting organizations.

Joint Commission Diagnostic Imaging Standards require training on burn risks, but do not explicitly require the preventative actions. ACR (Breast) MRI accreditation standards require the existence of a policy on “thermal burns,” but no specific content requirements and no explicit requirement for the preventative actions.

Projectiles

Similarly, with respect to magnetic projectile preventions, none of the major accrediting organizations have the explicit preventative steps that stop the overwhelming majority of projectile injury accidents (94% & 69%).

Ironically, the organization that comes close to invoking the site requirement for the ACR 4-zone model of suite access control and supervision is not the ACR, but The Joint Commission whose Diagnostic Imaging Standard functionally describes Zones 3 and 4. Similar to the organizations’ stances on burn prevention, TJC requires training on projectile risks and ACR requires the existence of a policy on screening (no policy content requirements). Neither of the two organizations explicitly require these established projectile preventative actions.

Hearing Damage

As with the #1 and #2 sources of MRI-classified injury, none of the major US accrediting organizations specifically require the use of hearing protection. Similar to the previous two injury hazards, TJC requires personnel to be trained on hearing damage risks, and ACR requires the existence of policy on “acoustic noise” (though, again, no policy content requirements). Again, neither of the two lead accrediting organizations explicitly require these hearing damage preventative actions.

Macro

I argue that we can measure the degree to which accrediting organizations fulfill their promise of assuring “the highest level of… patient safety” by looking at the overall adverse event rates for MRI. As long as the trending rates of accidents continues to climb at nearly three times the rate of MRI imaging, that’s a strong indication that accreditation programs, broadly, are failing to live up to their mandate to assure safety.

Micro

But of course it would be unfair to attribute declines in national MRI safety – across the board – to all of the accrediting organizations (when they each only have fractional slices of the whole market) without considering them individually. But when you do look at them individually, and measure the degree to which the specific provisions of their accreditation regimes effectively mitigate the most frequent sources of MRI patient injury, none of them appear to be doing a good job.

QC’ing MRI Safety Standards

If we want to apply quality control (QC) measures to MRI safety standards for accreditation, we should routinely be looking at reports of MRI injury accidents, both individually and in the aggregate, to identify the degree to which the organization’s safety standards successfully (or fail to) prevent injury accidents.

The inability of standards to stem the growth of accidents, broadly, or interrupt the causative pathways of individual MRI injury accidents, specifically, is an indictment of MRI safety standards across the industry.

 

---

Sources / Resources

1. US FDA MedWatch adverse event reporting criteria & reporting tools
2. US FDA MAUDE database search tool for MedWatch adverse event reports (use product code LNH to restrict search results to MRI-classified adverse events)
3. Article on comparisons between changes in US MRI exam volume and MRI adverse event volume
4. Recording of 2012 RSNA presentation on frequency, types, severity, trends & preventions of MRI injuries
5. Abstract of 2012 RSNA scientific paper presentation on frequency, types, severity, trends & preventions of MRI injuries
6. Metrasens 2018 infographic on the repeat of the 2012 RSNA-presented research with more contemporary data
7. “MRI-related FDA adverse event reports: A 10-yr review”
8. ACR MRI Accreditation Program
9. DNV Hospital Accreditation 
10. TJC Diagnostic Imaging Standards MRI elements
11. IAC MRI Accreditation Program