Alarmingly, The More We Know About MRI, The More We Seem To Hurt People

MRI has had one of the single most effective PR campaigns, ever. For more than 30 years we've contrasted MRI's absence of ionizing radiation with the potential harm that may arise from X-rays and CT scans, which use ionizing radiation (ionizing radiation is the form that damages DNA and can lead to the development of cancers). We've long called MRI "the safe modality," so it's a particularly cruel irony that the success of our marketing appears to have had the effect of dis-incentivizing the safety steps that MRI does require. Said plainly, our own PR is making MRI more dangerous.

The Risks

Let's begin with a straight-up confirmation... MRI does *not* use ionizing radiation. It uses magnetism instead ('always on' static magnetic fields, overlaid with time-varying magnetic fields), which means that -to the best of our knowledge- there is no known risk from MRI exams causing cancer.

But don't kid yourself... just because MRI doesn't share the most prevalent risk found elsewhere in a radiology department doesn't mean that it is risk-free. In fact MRI has a number of risks that are peculiar to it as an imaging technology; risks that don't have good analogues elsewhere in radiology or nuclear medicine. This means that MRI is different both in its technology and in the risks that arise from this technology.

The three most frequently-identified direct risks to MRI patient safety arise from the three different forms of the magnetic energies from the machine used in creating the medical images. The static magnetic field twists and pulls magnetic materials (the aptly-named 'missile effect'), the imaging gradients can induce a type of electrical current that can cause direct neuromuscular stimulation (most frequently seen as peripheral nerve stimulation, or PNS), and the radio frequency (RF) oscillated magnetic field can induce a type of electrical current that causes heating (both relatively benign diffuse heating, and the more worrisome potential for focal heating). Additionally, the mechanics of creating the oscillating imaging gradients can also create very loud noises, which may be damaging to the hearing of anyone in the MRI scanner room during the study.

So, while MRI doesn't use ionizing radiation, it can (and has) impaled people, crushed people, suffocated people, electro-stimulated people, heated people, burned people, and deafened people. So remind me again about how MRI is "the safe modality" when it can do all of these things to patients or caregivers? And all of this is just the MRI machine's / exam's direct effect on flesh-and-bone people... not even taking into account the potential of MRI's to interact with implants and devices that may also be present.

In addition to the above, there is also the potential that the MRI scanner can interact -in unwelcome ways- with implants, medical devices, and foreign bodies. These interactions have caused pacemakers to race a patient's heart, or not fire at all when the patient's heart needed it to. They have induced concentrated heating at the tip of a deep-brain stimulator lead and killed a few cm of brain tissue. They have caused malfunctions of implanted / onplanted medication pumps, dumping dangerous boluses of medication into their patients. They have twisted cerebral aneurysm clips, tearing vessels within a patient's brain. These 2nd order interactions with medical devices, that are not specifically designed to 'play nice' with MRI, are an entirely separate (and poorly quantified) set of risks.

The Safety

It may seem counter-intuitive, having just laid out all of the risks that MRI presents and potentials for gruesome injuries, to then follow up with the statement that MRI is profoundly safe, but it is... when we follow best practices. However, when everyone is confronted at every turn with the bumper-sticker slogan that MRI is 'the safe modality,' what does that do to the desire to intervene to enact safe MRI practices? If it's already 'the safe modality,' are limited resources really most-wisely invested in MRI safety?

I believe that this is the fundamental truth of MRI safety, today: the bumper-sticker slogan is so enormously successful that efforts to strengthen MRI safety are disparaged or actively undermined. I can't tell you the number of times that calls to follow best practices have been swatted back with, "but that's only a recommendation, not a requirement." The clear implication being, 'you're safe enough.' After all, MRI is 'the safe modality,' right?

The Trend

So you may be asking, "why all this hand-wringing?" Because of the direction in which we've been heading for quite some time.

Since 2001 (the Colombini Accident), there hasn't been a high-profile MRI accident in the United States that has similarly captured popular attention (there have been *lots* of MRI accidents and injuries, some fatal, but they've never garnered national press). Since that time, however, the rates of reported MRI accidents have risen, steadily and alarmingly, even when accounting for the rate of increase in MRI exams.

The following graph is one that I updated just a few months ago (you may have seen earlier versions... I've been tracking this data and producing graphics of it for more than a decade), and it shows the MRI-classified adverse events as taken from the US FDA's MAUDE database (shown in red), plotted against US MRI imaging procedure volume as released (or interpolated between years of published figures) from IMV (shown in blue). Because MRI exams are in the tens-of-millions, and reported MRI accidents are in the hundreds, simply comparing the numbers between the two would not effectively depict the comparative trends, so the chart below begins with the year 2000 (the year prior to the Colombini accident) and plots the percentage change in both the adverse event rate and the number of MRI procedures from the year 2000.

A few notes about the above graph. First, the data behind this graph is all in the public domain, and you are able to 'check my math' by going directly to the data sources yourself. Second, the more aggressive fluctuations in the red (adverse event) line is a result of a much smaller number of annual reported adverse events. A change of 100 adverse events is a massive fluctuation in the red line, whereas a change of a million MRI exams is a subtle shift in the blue line. Third, the red line only captures adverse event reports that were classified as MRI equipment adverse events (filed under product code LNH). The majority of medical implant or device failures that occur as a direct result of interactions with an MRI are not cross-classified as an MRI adverse event... they are listed exclusively under the implant or device. Quite simply this means that the actual number of MRI adverse events, as they are popularly understood, is significantly higher than the red line in this graph suggests.

The presentation of this data has made more than a few people upset, I think because it directly contradicts our bumper-sticker slogan. It has been offered that the increase (over time) in the adverse event curve is a result of better adverse event reporting (though it should be noted that not the FDA, ACR, TJC, ASRT, ISMRM, SMRT, or any of the MR system manufacturers have had broad campaigns in the past twenty years of which I'm aware to focus any attention on MRI adverse event reporting). While it is indeed possible that reporting has improved, I find such an explanation unlikely in the absence of any campaign to raise awareness of MRI providers about why, when, or how to report. Analogizing, how much more likely are you, today as compared to ten years ago, to report to the Federal Communications Commission (FCC) interference with your radio reception or poor wifi? If you don't know that it's even available as an option, or what it would do for you, how much more likely are you to report it, today, than back then? That's similar to where we are with MRI.

Additionally, some have sought to discount the mere attempt to track adverse events over time with the FDA data, indicating (correctly) that the FDA's adverse event data is grossly under-representative of what happens in the real world (by my own modeling, it may capture less than 2% of the actual number of MRI adverse events), and (also correctly) that the FDA admonishes against using this data to model trends (though the FDA's own people have ignored this admonition and published MRI adverse event trend information from this data). The good news about the inaccuracy of the FDA's data is that, in the past twenty years, there has been no substantive attempt to improve reporting of MRI adverse events to the FDA, meaning that the data is likely similarly-inaccurate today as it was twenty years ago (yay, consistency!). As long as the FDA data from each year is roughly equally under-counted, then a year-over-year analysis of the change is a valid dashboard of how we're doing.

Why Do I Think MRI Accidents Are Trending Upward?

If you're willing to accept (even temporarily) my argument that the numbers aren't going up as a result of better reporting, how do I support the conclusion that they're going up because of increased adverse events?

The way I think about it, MRI safety has three constituent parts that are required for risk and accident reduction: clinical safety (prescribing, protocoling, administering the exam), operational safety (training, screening, restricting access), and physical environment safety (4-zone, access controls, MR Conditional equipment). Let's look at examples within each of these, individually.

Over the past twenty years, what clinical generalizations can we make about MRI patients? Demographically, the US population is greyer and heavier (greater co-morbitidies) than it was in 2000. MRI has gone from all 'walkie-talkie, knees-and-heads,' to include ICU patients, ED patients, image-guided biopsies, and patients with all manner of implants and devices. In short, our patient cohort in 2020 presents increased risk as compared to the MRI patient cohort from the year 2000.

Now let's to the same with operations. How does the financial picture of performing MR studies in 2020 compare with 2000? Sites are getting substantially lower reimbursements for each study (even though capital costs and operational costs have gone up), resulting in pressures to decrease costs (often through staffing reductions) and increase throughput (more exams per day). 'Do more with less' has costs, and one of them is often safety.

Let's also look at the physical environment. While there are lots of parts of this that may reduce risk, let's just look at the main risk-introducing part... the MRI scanner, itself. Again, comparing with 2000, where are we today with field strengths, magnetic spatial gradients, RF power, gradient slew rates? All significantly higher today than the magnets of 2000. Virtually every MRI scanner technological improvement that improves imaging also carries with it increased risks. We are, today, doing a much better job with requirements for 4-zones and FMDs (for new or renovated MRI suites), but there's still a large number of 'grandfathered' suites still in use that have access to scanner rooms directly off of shared radiology corridors or offer no form of physical screening for ferromagnetic 'contraband.'

I think that the risks of MRI have inched upwards, on multiple fronts, since it was popularized as a clinical modality. I suppose there's a small comfort to be found in the fact that I don't think we're objectively worse at MRI safety than we were twenty years ago... but we have been standing still, not improving, while the risks have been growing slowly -but consistently- over time. Now we find ourselves having lost a lot of ground, and needing to make it up.

The Upshot

In the analogy of the a frog in a warming pot of water, we are the frog. The story goes that if you just increase the temperature of the water slowly, a frog in the pot will never recognize the danger and will stay in until it's cooked. As an industry, we've failed to recognize the increasing heat in the form of growing MRI risk, and our patients and caregivers are increasingly getting injured because of our failure to see what's happening around us, and respond accordingly.

We need to reverse the complacency that has built up as a consequence of our dangerously-effective 'MRI is the safe modality' PR campaign. We need to re-frame MRI risks as being different from those of ionizing radiation, and still equally as important to address. We need something akin to a MRI version of 'Image Gently' or 'Image Wisely' to ingrain into our culture the value of MRI safe practices. We need organizations and institutions like regulators, accrediting bodies, licensure organizations, and professional societies to support the normalization of best practices for MRI safety.

There is nothing in the overall trend in MRI accidents that we are powerless to improve. In fact, scroll up and look again at the red-line data for 2003 and 2004, and then come back here. I'm not going anywhere...

In the data, 2003 accident data (red line) matched the imaging volume data (blue line) in terms of percentage change, and in 2004 accident line dips below the imaging volume line for the one-and-only time in this twenty-year chart. Why do I think 2003 and 2004 saw these reductions that got down to / below the imaging line? The Colombini accident (July, 2001) galvanized the industry's attention on MRI safety (RSNA coincidentally had a categorical focus on MRI safety at their annual meeting that same winter), but it wasn't until a year later that there were industry best practice MRI safety standards in the form of the original ACR White Paper on MRI Safety (published June, 2002). The following year, 2003, was the first full calendar year that the US had both an acute attention to MRI safety and the best practice guidance to follow. Both 2003 and 2004 see reductions in MRI accident rates (2004 even dropping below the imaging curve).

Today we have even more (and even better) best practice guidance on MRI safety. The 'white paper' has undergone three subsequent revisions / expansions (and the fourth is due out at any time). So if we have better guidance and are still doing worse, then the key difference today is that we're missing that first of the two ingredients, acute attention to MRI safety. We, as an industry, have all of the tools we need to make significant improvements in MRI safety, we only lack the impetus to put them to use.

Until we re-find that impetus, we will continue to needlessly jeopardize MRI patients and caregivers while thumbing our noses at the best practices (that we already have) which could keep everyone safe.