A Game-Changing Statistic: 1 in 250

Although the medical profession has been harming unlucky patients for centuries, the patient safety movement didn’t take flight until 1999, when the Institute of Medicine published its seminal report, To Err is Human. And that report would have ended up as just another doorstop if not for its estimate that 44,000-98,000 Americans each year die from medical mistakes, the equivalent of a jumbo jet crashing each day.

Come to think of it, the quality movement also gelled after the publication of Beth McGlynn’s 2003 NEJM study, which produced its own statistical blockbuster: American medical care comports with evidence-based practice 54% of the time, a number close enough to a coin flip to be unforgettably disturbing.

These two examples demonstrate the unique power of a memorable statistic to catalyze a movement.

Last month, my colleague Rebecca Smith-Bindman, professor of radiology, epidemiology, and ob/gyn at UCSF and one of the nation’s experts in the risks of radiographs, gave Medical Grand Rounds at UCSF. Her talk was brimming with amazing statistics, but this is the one that took my breath away:

A 20-year old woman who gets an abdominal-pelvic CT scan (i.e., just about any young woman coming to the ED with belly pain) has a 1 in 250 chance of getting cancer from that single scan.

Did that fully register? One CAT scan, which until recently most of us ordered with no more restraint than we exhibit when asking the Starbucks barista for a tall latte, will cause cancer in one out of every 250 patients. Two-hundred fifty: that’s the number of students in my college Bio 101 class. Wow.

This is particularly scary given the remarkable increase in the use of this technology. Get this:

  • Three million CT scans were performed in the U.S. in 1980. In 2011, there will be 72 million, an average of 19,500 every day.
  • One in five Americans will receive a CT scan in any given year; some experts suggest that at least one-third of those scans are unnecessary.
  • Between 2000 and 2005, Medicare spending for imaging studies more than doubled, from $6.6 billion to $13.7 billion, twice the rate of growth of physician fees.

And, although none of these examples has quite the impact of the 1-in-250 statistic, there are lots of other scary risk data, such as:

  • The best estimates are that radiation from CT scans causes 29,000 excess cancers each year in the U.S., mostly in women.
  • Researchers estimate that 15,000 people will die from the direct effects of the 72 million CT scans performed in 2007 alone.
  • A 2004 study found that less than 50 percent of radiologists, and 9 percent of ER docs, were aware that CT scans could increase the subsequent risk of cancer.
  • A multiphase abdominal/pelvic CT scan has the same radiation wallop as 500 transcontinental flights, 450 chest radiographs, and 74 mammograms.
  • And those airport body scanners you’ve been so worried about? You’d need to be scanned 200,000 times in order to accumulate the radiation that you get from a single CT scan. I’m a 1K United flyer, but I won’t close in on 200,000 scans for the next couple of centuries.

In her grand rounds, Rebecca walked us through the multiple lines of evidence on the risks of radiation from CT scans, particularly those drawn from studies of Japanese A-bomb survivors and individuals who received radiation for both malignant (i.e., lymphoma) and non-malignant (i.e., acne) disease. All pointed to the conclusion that doses in the range of those delivered by CT scans are fully capable of causing cancer.

Remarkably, with all the attention given to regulating food and drugs, the radiation delivered by CT scanners has gone largely unregulated. (If you ask me, I’d rather receive a precise and predictable dose of radiation than of Vitamin D or Azithromycin.) Rebecca found that CT scanners at four Bay area hospitals delivered radiation doses 66% higher than the usually-quoted doses, and that there were staggering variations (up to 13-fold) among different scanners performing precisely the same test. In her talk, she blamed the lax regulations on radiation physicists, fastidious types who have been reluctant to take a stand on maximum radiation doses since they can’t define those doses precisely.

While I’m sure that’s true, I have to believe that some of the reluctance to blow the whistle can be traced to the usual Medical-Industrial Complex: scanning equipment manufacturers, radiologists, and hospitals who have no particular interest in killing this particular egg-laying goose. If you doubt that these forces are at play, witness the billboards for $1000 total body scans that line Florida’s highways (scans that, when performed in healthy people searching for asymptomatic tumors, undoubtedly cause more cancers than they cure). Even now, despite powerful evidence of the risks, there are some in the radiology community who don’t find the science compelling enough to alter their practice. The parallels to the Global Warming debate are eerie, and troubling.

Even if the risks turn out to be less than we fear, most skeptics now agree that we’re causing a lot of cancers, and that many could be prevented if we took a few sensible steps. Manufacturers, hospitals, and radiology facilities should test the radiation exposure of their scanners, with the goal of decreasing the variation and delivering the minimum dose that creates an acceptable image. Ultrasounds should be substituted for CTs when possible, such as in follow-up of patients with documented kidney stones. There is evidence from Mass General that the use of computerized appropriateness protocols can markedly cut down on the number of CT scans, and thus the cancer risk. And, if we need to obtain the patient’s informed consent before transfusing a unit of blood, we should also do so before ordering a CT scan, since the latter is a far riskier procedure.

But changing culture will be more important, and harder, than changing protocols. We physicians have become so accustomed to saying “Get the scan” that we have turned our brains off. Several months ago, I cared for a woman with a painful lumbar compression fracture of unknown duration. We asked the orthopedic surgery service to see her in consultation, and the resident’s recommendation – made without a hint of self-awareness or irony – was that we obtain both a CT and an MRI. I was dumbfounded. Yes, each test can provide slightly different information, but I don’t believe that both were absolutely necessary; nor did a couple of experts I later spoke with. (We ended up getting the MRI only, which produced all the information we needed.) Somehow, we must find a way to break our reflexive radiographic profligacy.

As we struggle as a nation to “bend the cost curve” and we grapple with the nexus of low yield and expensive medicine (the dreaded “R word”), let us all agree that when we have an issue like this – an overused technology that harms or kills thousands of patients each year – we come together to do the right thing. CT scans can be immensely helpful, even miraculous, at times, but there is no question that the right thing is to Just Say No far more often than we ever have before.

17 Responses to “A Game-Changing Statistic: 1 in 250”

  1. cory February 11, 2011 at 2:16 pm #

    It is not clear whether the 1 in 250 is an actual number or an estimate. Big difference.

  2. Bob Wachter February 11, 2011 at 6:41 pm #

    Thanks, Cory – the one in 250 number is an estimate, largely drawn from the known risk of cancer seen in people who received similar doses of radiation in different contexts, such as people in Hiroshima who were relatively far away from Ground Zero. Having read over the data and the sources (all of which I’ve linked to in the blog), I find the estimates credible.

    But the point is, even if the risk estimates are off by a factor of two, this is still a far higher risk than any of us previously appreciated and it should be factored into the way we think about and manage this technology. Do you disagree?

    The debate reminds me of the early years of the patient safety field, after the IOM estimate of 44,000-98,000 deaths per year came out and created such a buzz. Some people pushed back on these numbers, which also were estimates drawn from credible background research. My point then was that, even if it was not a “jumbo jet a day” worth of deaths from medical mistakes but “merely” a Greyhound bus-worth of deaths, this was still a very big problem that we had been ignoring. The same is true here.  

  3. Bob February 11, 2011 at 7:25 pm #

    The medical industry has traditionally been exempt from the radiation exposure limits and radioactive contamination standards required of every other industry. This stems from two fundamental historical principles, a) “the doctor knows best” and b) “this mess will cost too much to clean up.”

    While there’s a world of difference between protecting workers from occupational radiation exposure and using radiation as a diagnostic and therapeutic tool, it sounds like the medical industry needs to take a page from the nuclear power community regarding the concept of ALARA – “As Low As Reasonably Achievable.” Indeed, WHO and the ICRP have been instrumental in effecting change in this area, but it’s telling that movement had to come from those agencies rather than insurers, regulatory bodies, or the medical community.

    Remember, we’re considering the damage from typical therapeutic and diagnostic use of radiation, not accidents like losing sources in people’s noses (happens occasionally when treating septum cancers) or device faults such as the infamous case of the Therac-25. Medical use of radiation is the largest source of artificial radiation humans are exposed to; it would be interesting to see the trend in the average dose received by patients over the past 60 years.

    Two questions: how much of this radiation exposure is due to ‘CYA’ medicine (“run another test, just to be sure & so we won’t be sued later”)? And how much could be reduced by good medical records portability between different specialists & facilities?

    Anecdotally, I have a fair amount of blood work done at a large teaching hospital for several doctors (endocrinologist, GP, rheumatologist, etc.) and being all on the same records management system, they get to see the results of each others’ tests, and add their own tests whenever I have blood drawn. This tends to cut down on the total number of blood draws, a sort of economy of phlebotomological scale. As a patient, I appreciate being poked less while still providing each of my doctors with the data they want. Surely improving access to records is a way to keep from redundantly irradiating people, at least for basic diagnostic procedures.

  4. cory February 11, 2011 at 9:57 pm #

    If we stay on the issue of danger of CT scan, not whether we are getting too many or preventable hospital deaths, which are both long involved subjects on their own, I still don’t see this.
    The article says the abdominal CT is roughly 31 mSV (unless I have the wrong type of scan) and a chest X-ray is in the range of 10 mSv so where did the 450 number times as much as a chest radiograph come from? Are there really CT scans giving 4500 mSvs? I don’t know but if there are that’s a problem.
    (Background radiation is about 1-2 mSv). Look there is a lot of radiation going around, no question and we would do well to prevent what we can but a number like 1/250 people getting cancer form a CT scan in the absence of true epidemiologic data, does not advance our knowledge. Using the Hiroshima example we would have thousands of people coming down with these cancers each year. CT scans have been in operation for a generation, but I can’t believe we are seeing this. Perhaps we will in the future but it raises the question of why we haven’t so far (or if we have why no one has come close to documenting it).
    This is a very difficult long-term issue, not readily explained with our current info. IT would be better to do some long-term epidemiologic studies and to make sure every patient has a record of roughly how much exposure they have had, a la vaccination, rather than to bandy about extremely hard to believe extrapolations.

  5. Mike February 12, 2011 at 8:34 am #

    Cory – I think you’ve got a units mixup; the number I’ve seen quoted for a CXR is ten milli*rem*, not ten milli*Sieverts*. 10mRem = 0.1mSv, so 450 times -> 45mSv which is in the same ballpark is the 31mSv (3.1 Rem for those of us, like me, who still think in archaic terms) that you – and the article referenced above – quote for the CT. Which, randomly enough, is almost at the 5 Rem that the NRC used a few years ago for the annual dose limit for radiation workers (probably still the same now, but I haven’t done any nuclear stuff since grad school). So if a radiation worker shows up as a trauma patient and gets pan-scanned, they’re probably done with radiation work for the year… [grin] (not actually sure if medical radiation counts towards occupational limits) Side note – if I’m interpreting what that study said correctly, the 31mSv was for a *multiphase* CT A/P, implying that a single-phase scan like most of the ones we do in the ED should be a fraction (1/2 – 1/4, depending on how many phases their multiphase scan was) of that dose.

    I’ve got my doubts about the quality of the data behind BEIR-7 (which is where everyone gets their risk estimates from), as they did a bit of picking-and-choosing. Whether or not it was justified is an excellent question – they tried to pick the higher-quality evidence, but there just isn’t much actual high-quality stuff out there – but a *lot* of BEIR-7 is based on the Hiroshima and Nagasaki data and it just isn’t that robust (the dose estimates, for example, are kind of a guess). There are some other datasets that show a much smaller effect of ionizing radiation, but they weren’t used much due to concerns for bias (for example, I’m blanking on the name but there was a study of guys who worked in a shipyard on nuclear subs…and had no increased incidence of anything despite increased radiation exposure…but it was written off as “the healthy worker effect” i.e. the group being healthier at baseline than the comparison group). I suspect that the 1:250 risk is somewhat significantly overstated, but I freely admit that I base that on a slightly different reading of the same literature and have no hard evidence to support me.

    Having said that, as an ED type myself, I (and most of my colleagues) do try hard to avoid scans particularly in younger people – the same belly pain that I’d scan a 60-year-old for without thinking twice I’ll try hard to avoid scanning in a 20-year-old – because whatever carcinogenic effect may result from ionizing radiation, it’s almost certainly going to be magnified the longer the patient has left in life to manifest it. Not to mention the more direct effects on little kids’ CNS – what seems to be a pretty likely, albeit small, cognitive hit from scanning a 6 month old’s brain scares me a lot more than the risk of radio-induced cancer down the road.

    -Mike, ED doc (despite an undergrad degree in nuclear engineering)

  6. Cory February 12, 2011 at 9:13 pm #

    Mike:
    Thanks for clearing that up.
    Mea culpa

  7. Bob Wachter February 12, 2011 at 11:18 pm #

    This comment comes from e-Patient Dave (again, folks, sorry if you’re unable to post comments — we’ll be changing platforms soon and all will be well. For now, just email me with any comment you’re unable to post):

    Would this qualify as an urgent public health issue, with bulletins being issued forth to every office? Should this be an added index card on every informed consent signature?

    I understand Cory’s concerns, but I don’t think citing To Err Is Human is off topic; that was about our resistance to realizing we cause unexpected harm far more often than we would think. This is an enormous obstacle to change. (Who wants to change something that they think isn’t happening?)

    I’ve wondered how it is that good smart people can have a blind spot like this, and I think it’s because accidental medical killings happen in microevents (one at a time, scattered all over creation), not by the busload or planeload, so there’s never any massive event.

    But that resistance is EXTREMELY unscientific, and leads on to greater harm: as the Inspector General’s report in November showed, if anything things have gotten worse since To Err: the IG’s retrospective analysis of Medicare data showed a rate of accidental killings equivalent to 15,000 a month (not year), and that’s just among Medicare patients.

    So I assert that the issue of our cultural blindness is indeed relevant as we ask ourselves “What?? Could this really be happening?? It doesn’t SEEM likely.” To the contrary, it seems callous (unintentionally I’m sure) to say “Let’s not get all cautious about this – let’s watch for a decade or so and then see how many of the ladies have actually gotten cancer.”  That’s not what I want for MY daughter…

  8. Brian Clay, MD February 13, 2011 at 4:42 am #

    Two things:

    First: as electronic medical records advance and sharing of medical data becomes easier, I am hopeful that at least redundant scanning due to lack of outside study reports will be significantly reduced.

    Second, a comment Beth’s McGlynn’s study and the coin flip analogy: it really isn’t the right analogy, and I think it gets made frequently because the percentage of compliance with evidence-based care cited in the study happens to be so close to 50%. In fact, the analogy should be to some metric of a skill level (think batting average, free throw completion rate, etc.), rather than something derived from a random event. This isn’t to make Beth’s finding more palatable; in fact, it makes it scarier: if we KNOW the evidence, and we KNOW what we should be doing in certain clinical situations, then our patients should reasonably expect us to do it right nearly 100 percent of the time. 54% seems woefully short, although I would like to see a re-do of the study now that it’s almost 10 years later.

  9. Cory February 13, 2011 at 5:38 am #

    I do not mean to minimize problems like CT caused cancer or iatrogenic deaths.
    What I do mean to do is put them in proper perspective.
    Dr. Wachter is a baseball fan- he has written about baseball and medicine. AS in baseball, quoting numbers that you think might be true is different from measuring and analyzing.
    Example -“Alex Rodriguez might get to two balls a game another third baseman doesn’t get to”. Sounds like it could be true – people thought stuff like that was true for years, until they actually measured it and came up with quite different numbers (the difference between the best and worst third basemen in baseball is not two balls a game). Measure it with good techniques and then describe it.

    The IG report is patent nonsense if it is claiming 15,000/month in Medicare patients. I say that without reservation. Do you know what that number means?
    180,000 iatrogenic deaths in Medicare patients per year.
    Answer this- there are roughly 2-2.5 million total deaths in the United States every year. That includes everything, accidents, homicides, in hospital out of hospital, terminal patients, nonterminal patients, Medicare, non Medicare. The IG is telling me that nearly 1 in 10 deaths is due to iatrogenic causes? And it is probably greater than 1 in 10 if we exclude out of hospital deaths for unpreventable things.
    And it is much greater than that if as suggested this is only Medicare patients.
    Forget even all the terminal patients who die in hospitals or nursing homes where death is expected and error is not an issue.
    One out of say, every eight deaths in the entire country is due to medical error? I would like to see someone prospectively document that – it wouldn’t take long if it were true.
    Every hospital in the country would be seeing at least one iatrogenic death per week, some many more than one. You wouldn’t need alerts- it would be patently obvious to everyone, the doctors, nurses, families, ward secretaries. We would be doing nothing but M and Ms and testifying in lawsuits.It is literally impossible that that many people are dying of iatrogenic causes.
    Point being- not that there isn’t a problem but scare tactics and wildly extrapolated numbers are no substitute for reasoned analysis. There is a problem with iatrogenic deaths, there may well be a problem with CT caused cancers- but let’s not just go throwing out wild numbers -it ultimately hurts our credibility as a profession. And it makes the real numbers less believable when we do have them.

  10. AD February 13, 2011 at 6:57 am #

    The arguments above are largely academic. The “screening CT”has replaced the need for a history and physical.In our ED it is almost part of every work up together with a troponin and BNP ordered for a hang nail or a real medical emergency.
    The risks may be debatable as one argues estimates and realities.
    The $$$$ are real, the receiving patient and ordering physician should have skin in game the results may be surprising.

  11. Mt Doc February 14, 2011 at 7:56 pm #

    I don’t doubt there’s a problem and think it might be a good idea for patients to sign an informed consent that radiation is hazardous – this might do more than anything to limit scanning – but would suggest that the number of “unnecessary” scans is debatable and also would argue that the real risk is unclear. It’s interesting that the example used in the article to illustrate this supposedly very common problem occurred “several months ago”, the scans were ordered by a resident still in training rather than by a fully trained MD, and a scan (the MRI) was still done. Not a good way to illustrate a problem which is being portrayed as so universal.

  12. JustAPatient February 15, 2011 at 1:33 am #

    “Rebecca found that CT scanners at four Bay area hospitals delivered radiation doses 66% higher than the usually-quoted doses, and that there were staggering variations (up to 13-fold) among different scanners performing precisely the same test.”

    That statement alone is unconscionable! Regardless of how accurate the statistics quoted in Bob’s article are, or how variable the relative impact on cancer rates are for these scans, it is undeniably made worse by this. And you all in the field can and must do something about that. To use the airline analogy, that margin of error would make getting through security a breeze, because no one would ever get on a plane.

  13. medped February 19, 2011 at 10:21 pm #

    What you are not saying is why do ED physicians order a CT for every kid with a belly ache? Not for financial gain. Not to speed up the ED through put. The elephant in the room here is the tort system. Order the CT and the (cancer) risk is put on the patient,
    use good clinical judgement but miss something, and the risk is all on the physician.
    Develop treatment protocols that if followed protect physicians from liability and you will see the number of scans decline.

  14. rafmar February 22, 2011 at 5:39 am #

    Medped is spot on. There is no mystery here; only an inconvenient truth for the supporters of Obamacare. The overwhelming proximal cause of excess imaging in US medicine is litigation. Tort reform is the only solution, but the Dems are too beholden to the trial lawyers to allow this to happen.

  15. Shree February 26, 2011 at 12:35 am #

    How do we reconcile Value with, “I don’t want to miss something for fear of getting sued”? I think tort reform, informed consent, and standard protocols for tests are needed. We should be more thoughtful of the tests we order and use a heightened sense of clinical judgement when treating our patients.

  16. Proud RN March 16, 2011 at 5:40 pm #

    Interresting to read this blog now, after the earthquake in Japan and the issues with the nuclear power plants

  17. MURRAY WACHTER October 8, 2011 at 4:05 pm #

    Hi, Will see you next week. Just finished reading your blog, understood about 30%. How about some insight into older folks problems. You will see plenty of that here in Boca.
    Mainly, we past 80 find remembering, a full time job. Last week I rushed to an appointment at my Dermatologist. Arriving back home I was greeted by my lovely wife, your Mother, “You just took my appointment.” Playing tennis 7 days a week, we find our biggest problem is remembering scores, sometimes coming close to blows. Using aftershave for mouth wash has become quite common. The interesting thing is that we all have become experts in medical treatments, mainly drug usage. Some guys can go on for hours explaining the difference between statins, zine, oxcine and shmine. The computer does not help, it only scares us to death.
    We will talk more during dinner.
    Love–Dad

Leave a Reply