CT Overuse: The Problem Beneath the Problem


The headlines scream that emergency physicians overuse CT scans, but cost and radiation are only part of the issue at hand. The most enduring CT challenge will be practice variability across the nation.

Even the most casual review of the medical literature finds article after article indicating that we are doing too many CT scans. The number of tests being ordered is even beginning to make radiologists squirm – and these folks make their living reading these studies.

When you look for culprits, emergency providers appear at the top of the hit list for obvious reasons. What other group of physicians sees 2.5 new patients per hour? What other group sees “emergencies” as their primary patient type?

According to a 2012 survey of about 1,200 hospitals conducted by the Emergency Department Benchmarking Alliance (EDBA), there were about 20 CTs done for every 100 ED patients. This number is close to the National Hospital Ambulatory Medical Care Survey (NHAMCS) (2010) results in which it is noted that 16% of ED patients received a CT scan (with a little less than half involving the head).

It seems like there are two major reasons people want the ED to cut back on CTs – cost and radiation. We’ll take each in turn. First, if we’re going to cast CTs as a financial burden on the system, we need to separate costs from charges. How much can a CT really cost to perform? My intuition is that the CT machine is an absolute printing press for money in the hospital. So wouldn’t the hospital be all in favor of ordering lots and lots of CTs? I never heard a hospital executive complaining that we are ordering too many CTs (except perhaps on the capitated payment patients).

When considering costs, there are all sorts of variables to consider – amortized cost of the CT scanner, maintenance, hospital overhead, tech costs, etc. Fortunately, CMS has done the work of calculating these costs for us and has determined, as a result, what it should pay after adding in some profit for the hospital. Most hospitals don’t agree that CMS is paying a fair amount and it is common to hear that they think CMS is covering 70% of costs.

Be that as it may, just for a matter of reference, let’s look at what CMS reimburses hospitals for CT scans to get an idea of the spread between hospital charges and CMS reimbursement. Here are data from a hospital system in Arizona.1 The charges are not particularly current, but they’re likely close enough to make the point: the gap between charges and what CMS pays is remarkable:

Average Charge: $777
Average Payment with Medicare: $186

Average Charge: $1,492
Average Payment with Medicare: $330

Average Charge: $1,272
Average Payment with Medicare: $326

Average Charge: $1,785
Average Payment with Medicare: $226

Average Charge: $979
Average Payment with Medicare: $565

So it appears that the huge charges for CT scans are, in fact, a multiple of what is considered fair payment by CMS. So perhaps the idea that CT scans are so costly is really not the case.

The second issue regarding the ordering of CT scans is the radiation. We typically hear that chest or abdominal CTs are equivalent to the radiation received from 500 chest x-rays and that it can be expected that one in 2,000 patients getting a chest or abdominal scan will die from the cancers that these scans will cause. Typically we read that the dose of radiation for these scans is in the range of 8-12 mSv.

But what if the radiation dose could be reduced to approximately a tenth of these doses? That’s just what the following paper says is feasible and, in the process, asks the question, “Is the plain film dead?” If true costs are much lower than charges and radiation is low, why not replace CT scans with plain films? The scans are quick and clearly superior to plain films in demonstrating pathology. Take a look at some of the examples in the following study from Vancouver General Hospital.2 Using a variety of techniques, they were able to markedly reduce radiation doses while still maintaining adequate image quality:

  • An ultra-low-dose CT of the cervical spine in a 28-year-old pregnant female was performed at an effective dose of 0.6mSv compared with 6mSV at 2008 levels (and 0.2mSv for the average plain C-spine x-ray in 2008).
  • The effective radiation dose for an ultra-low-dose chest CT was 0.52mSv, compared with 7mSv for a standard CT and 0.1mSv for an average chest x-ray performed in 2008.
  • The effective radiation dose for an ultra-low-dose coronary CT angiogram was 1mSv compared with average doses of 16mSv and 7mSv for an average coronary angiography and CT of the chest performed in 2008.
  • An ultra-low-dose CT performed to visualize a renal calculus had an effective radiation dose of 0.4mSv compared with average doses of 0.7mSv for an abdominal x-ray and 8mSv for a CT of the abdomen in 2008.
  • An ultra-low-dose abdominal CT for suspected appendicitis was associated with an effective radiation dose of 0.56mSv.

With radiation doses in some cases less than those with plain x-rays and still providing adequate quality and better resolution than plain films, it is conceivable that, with some time, the plain x-ray may go the way of the dodo bird – especially if charges become less of an issue (as in a system in which charges don’t matter – e.g., Kaiser Permanente).

So charges might be starting to approach costs, and radiation is on the decline. But one major problem with CT remains. And that is that physicians can’t agree on when to order them. This is the issue of practice variability, and it has been repeatedly demonstrated in the literature. There are several studies that focus on variability when it comes to ordering CTs for all manner of complaints. I absolutely admire the courage of the authors of these papers. They are shining the searchlight of scientific inquiry not onto others but on themselves and are demonstrating that, within their own groups, variability with regard to CT ordering is pretty astounding.

Data from the most recent study we have are from Virginia Tech.3 The paper looked at the use of CT ordering by 49 of their staff emergency physicians between 2008-2012 in their teaching hospital, trauma center ED (about 85,000 visits). Similarly to the numbers quoted in the EDBA and NHACMS databases, 23% of the patients had a CT.

  • Multiple scans were performed in 18%
  • At least three scans were performed in 5.6%.
  • As seen in other studies, frequency of CT ordering was not related to physician experience.
  • After adjustment for confounders there was almost a three-fold difference between the low and high orderers (11.5% vs 32.7%).
  • High head CT use was predictive of high CT use across the board (we all know the physician who never met a test he/she didn’t like).

With regard to variation in specific studies:

  • Abdomen (15-52%) (the variation was eight-fold in discharged abdominal pain patients)
  • Chest (4-32%)
  • Headache (17-76%)
  • Shortness of breath (4-29%)

The authors concluded that the variation was “dramatic” – no argument there.

Two additional studies looking at the practice of the physicians at Brigham and Women’s Hospital confirm the gross variation that occurs in ordering head CTs. Again, the courage of the authors in showing the world their dirty laundry is laudable. The first4 looked at the ordering of head CTs by 38 of their staff physicians. Unadjusted rates of scanning varied between 4.4% and a remarkable 16.9%. Even after adjustment for confounders the numbers were impressive (6.5% vs 13.5%). In the subset of patients with headaches the range was 15.2% vs 61.7% (21.2-60.1% after adjustment). Again, level of experience was not linked to the frequency of CT ordering (a very discouraging fact).

To make matters worse a second study at Brigham and Women’s looking at head CTs for trauma confirmed that head CT ordering was all over the dartboard. Frequency of ordering ranged from 7.2% to 24.5% with a remarkable outlier at 41.7%. Good Lord.

Finally a paper looking at variation in ordering of head CTs for pediatric trauma between 40 hospitals.5 But remember, hospitals don’t order CTs, clinicians do. Of nearly nine million pediatric visits, 1.8% involved children discharged after an evaluation for minor trauma while 0.07% had significant head injuries. The rate of CT imaging ranged between 19% and 58% with a median of 36%. I’m sure that, if looked at, individual variation at each hospital would also be most discouraging.

Addressing variation in physician practice is very, very difficult. Physicians aren’t interested in having their practice compared with colleagues, there are no tangible financial motivators to narrow variability (at least not now), and physicians will make all sorts of excuses for their behavior (inordinate fear of malpractice suits, concerns regarding patient satisfaction, risk aversiveness conveyed to them as residents, you name it). Fundamentally strong and committed leadership is needed to attack unjustified variation. There are some techniques that have been used to try and narrow variability.

  • Physicians need to be shown the numbers and the numbers cannot be contested (use large sample sizes of patients, try to adjust for any variation in schedules [some doctors work more nights than others, busier shifts than others, etc.]).
  • Try and pick areas where there is compelling literature to support your position.
  • Move the herd rather than trying to single out the bad apples.
  • Involve the clinicians in embracing those changes that you are trying to make (may be tough when it comes to ordering imaging).
  • Authority-based feedback is the ultimate tool (creating financial incentives to alter behavior – but it is not easy when imaging is involved).

The way I see it, when it comes to the more public problems with CT overuse – over-exposure to radiation and runaway costs – there is a light at the end of the tunnel. But just beneath the headlines we’re still wrestling with how to deal with gross variation in practice, and we’ll need to continue addressing this for a long time to come.


  1. http://www.jcl.com/hospitals/average-pricing-information/ct-scans
  2. (McLaughlin, P.D., et al, Can Assoc Radiol J, 64[4]:314, November 2013)
  3. (EMA 3/14 – #30, Variation in Use of All Types of Computed Tomography by Emergency Physicians, Levine, M.B., et al, Am J Emerg Med, 31[10]:1437, October 2013)
  4. Prevedello, L.M., et al (Am J Med 125[4]:356, April 2012
  5. Mannix, R., et al, J Ped 160[1]:136, January 2012)

Dr. Bukata is the Editor of Emergency Medical Abstracts.


  1. Thomas Benzoni on

    I can’t hold my breath this long!
    Did the variation yield patient-level benefit?
    Did the high orderers find stuff that mattered?

  2. Paul DelPorto on

    Until we are empowered to say no to patients (regardless of litigation, cost, et al), there will always be patients who demand the test. It would be interesting to see the above data and take out all the CTs that the physician didn’t want to do, but was requested by the patient.

Leave A Reply