Friday, June 24, 2016

Irrational fear of CT scans in appendicitis

By Skeptical Scalpel and Saurabh Jha*

Simple appendicitis cannot be distinguished from complicated appendicitis by clinical examination and laboratory findings say Finnish investigators. They looked at data from their randomized prospective trial of antibiotics vs. surgery for treatment of appendicitis and concluded that only CT scans could reliably differentiate the two entities.

The study involved adult patients from 18 to 60 years old; 368 of whom had uncomplicated acute appendicitis and 337 had complicated appendicitis—appendicolith, perforation, or abscess.

Duration of symptoms, C-reactive protein, white blood cell count, and temperature were significantly different between simple and complicated appendicitis patients. However substantial overlap of values meant they were not helpful in predicting the presence of complicated appendicitis.

Receiver operating curves for C-reactive protein and temperature areas under the curve do no exceed 0.77. Combining these parameters did not improve accuracy.

The paper concluded that CT scanning is essential in diagnosing acute appendicitis and identifying simple and complicated cases.

A companion study by many of the same authors looked at 1321 patients who presented with clinical and laboratory findings of possible appendicitis. Since their protocol called for confirmation of the diagnosis, all patients underwent CT scans, and 351 (27%) did not have appendicitis.

Just over half of that cohort had no abnormal findings, but another cause of their acute abdominal pain was found in 160 (46%) of patients including 45 with acute diverticulitis, 39 with gynecologic problems, and 76 with “other miscellaneous diagnoses.”

As in the other study, white blood cell count, C-reactive protein, temperature, and duration of symptoms were unable to discriminate patients with appendicitis from those having CT scans showing other diagnoses or normal findings.

Both Finnish papers do not mention the use of ultrasound for diagnosing appendicitis. If ultrasound is available and accurate in your hospital, use it, but we are not aware any research validating ultrasound as a method of distinguishing simple from complicated appendicitis.

In the discussion section of the paper, the authors cited a New England Journal of Medicine study showing that low-dose CT scanning delivered less than one-third the dose of standard CT scan while maintaining similar sensitivity and specificity.

But the dangers of radiation from standard-dose CT may be overestimated.

The cancers attributable to CT are projections based on the linear-no-threshold (LNT) hypothesis. LNT assumes that the body reacts linearly to radiation at all doses; meaning there’s no safe dose of radiation. Attributable cancers are an assumption based on an assumption–a quadratic assumption–the mother of all assumptions.

We know the rate of excess cancers in Hiroshima and Nagasaki atomic bomb survivors exposed to 1000 millisivert (msv) of radiation. We assume that the attributable risk of cancer from abdominal CT with a dose of 10 msv based on LNT is 1/100th the risk from 1000 msv. This assumption is unscientific because biological systems are non-linear. LNT ignores basic radiobiology–DNA repair.

Toxicologists have a maxim: “dose is poison.” 100 % oxygen causes blindness in neonates. Excessive water intake can cause cerebral edema. Would it not be absurd to claim there’s no safe dose of oxygen and water?

Radiation is everywhere. The annual background radiation is 3 msv (we won’t tell you that it is 1.5 msv higher in Colorado, because we’ll spoil your skiing vacation). In scenic Kerala, India, a popular tourist destination, annual radiation is 70 msv. Keralites receive the equivalent of 7 CT abdomens every year, and 700 chest x-rays a year or two chest x-rays a day for their entire lives. If LNT were true there should be 300,000 excess cancers in Kerala annually. The Indian government should evacuate Kerala stat. However, researchers have found no increase in cancer in coastal Kerala. I’ve (SJ) been to Kerala. I can attest people aren’t glowing and don’t have three eyes.

Ironically, the LNT was put forward because physicists thought models, other than linear models, would be too difficult to understand. Linear thinking is the easiest thinking. Work hard and you’ll do well; work harder and you’ll do better. It is a convenient, but wrong, heuristic about our non-linear world.

This does not mean we should eat radiation for breakfast. There is compelling evidence of carcinogenesis at doses greater than 100 msv. We support the judicious use of CT scans, but clinical acumen is insufficient if you are going to use antibiotics to treat simple appendicitis.

The “no safe dose of radiation” is a sensible precautionary principle–better safe than sorry. However, through sleight of hand, it is now asserted as fact. This is irresponsible.

Prudence requires we neither irradiate people frivolously nor scare people frivolously.

*Dr. Jha (@RogueRad) is an Assistant Professor of Radiology at the Hospital of the University of Pennsylvania


Vamsi Aribindi said...

Dr. Jha and Dr. Scalpel,

I'm convinced by your argument that radiation risk is non-linear. But there is an implicit assumption in your argument that CT Scans are on the right side of that non-linearity.

After all, what if it's not just non-linear based on absolute dose, but non-linear based on the rate of the dose as well? The annual background dose may be 3 msv, but that is spread out over a year. In contrast, the radiation from an abdominal CT is 10 msv over approximately 1 minute- the rates are different by 5-6 orders of magnitude. Logically, a few DNA breaks here and there over the course of a year have plenty of time to be repaired, while a million breaks over the course of a minute overwhelm the repair mechanism.

Vamsi Aribindi

Anonymous said...

Another perspective

saurabh jha said...


You are correct that the rate of dose of radiation is important. It is for this reason that cumulative dose (or how many CTs one has had), which is now tracked, is not terribly helpful.

Can 10 msv be the threshold for radiation-induced malignancy? It is not impossible but unlikely. The radiation-cancer curve is more likely to be sigmoid than linear. This means once the threshold is crossed there should be a rapid rise. Were 10 msv the threshold, 20-50 msv should be the part of the rapid rise. The extra cancers at doses > 100 msv are detectable. Below that, epidemiological studies are, at best, contradictory.

Saurabh Jha

Anonymous said...

Living at high altitude increases background radiation, which is continually delivered in minute increments. I see no reason to assume that a burst of radiation delivered in a fraction of a second is as safe as the same dosage spread out over a full year. There are also epidemiological studies showing that CT scans increase cancer risks, particularly in children. Of course in some circumstances the benefits exceed the risks. However, pretending that there are no risks will have two negative consequences: doctors who believe it will dish out needless CT scans that later cause cancer, and patients who don't believe it will refuse useful CT scans because there's no way to discuss risk-benefit ratios with someone who doesn't acknowledge one half of the equation.

saurabh jha said...

It seems we all agree that LNT is far too simplistic. Believe me, that’s a major agreement.

Epidemiological studies are a very poor measuring instrument to tease apart how much excess cancers are from radiation at doses typically seen in CT. Particularly as intensive imaging unmasks a reservoir of clinically silent cancers – a well-described phenomenon known as overdiagnosis.

It is important not to confuse “we do not know the risk therefore it is better to take precaution” with “radiation is giving us cancer.” It is also important not to confuse “be prudent with CT” with “don’t use CT scans at all.”

Prudence, not fear-mongering, is the way forward.


Korhomme said...

I'm a surgical dinosaur, brought up in the time when the diagnosis of appendicitis was entirely clinical. We relied on Zachary Cope, and his alter ego, Zeta.

And then came ultrasound, and it only seemed to delay diagnosis – if we actually got an answer from it. And now CT; well, in the NHS in my part of the Celtic wastelands, getting one was always a matter of being nice to a friendly radiologist.

Have we completely lost clinical skill? I don't mean the skill to guess that it might be appendicitis, I mean the skill to be (reasonably) certain. Is this really progress – or am I just superannuated?

Skeptical Scalpel said...

The boat has sailed. In my area, most patients with abdominal pain get CTs before surgery is called.

I blogged about this 6 years ago [] saying "In a non-teaching hospital where there are no residents, it is very difficult to have every patient with a suspicion of appendicitis seen by a surgeon. When the emergency physician calls and says she has a patient with a positive CT scan for appendicitis, the diagnosis is correct more than 95% of the time."

artiger said...

If, like me, you work in a hospital that does not have a surgery crew and anesthesia in house 24/7, one of the last things you feel like doing is wasting time up at the hospital while waiting for them to drive in and set up. So I don't mind if ER docs want to scan someone before calling me. Our ER docs are usually primary care guys who have quit private practice, and as such they don't have a good feel for appendicitis anyway. Patients in this day and age usually want to be scanned before they get cut, and I'd rather spend 30 minutes or so up there as opposed to a few hours.

I trained in the 1990's, and we had to examine a lot of bellyaches without the benefit of a quick CT. In all modesty, I think I got pretty good at making the diagnosis without a CT (more so on exam than history, and regardless of the WBC). I wasn't 95% accurate though. Still, it comes from seeing and examining a LOT of patients, and today there just isn't enough time for the volume of that kind of learning curve, neither in residency nor afterward.

Skeptical Scalpel said...

I agree. I had the same issue with an OR staff that was not in-house after 11 PM. It makes a huge difference to have positive CT scan so they can be on their way in and ready after you have seen the patient.

I know there are those who feel appendectomies can be delayed without harming the patients. I always felt like it was better to get it done at night instead of hoping for an opening in the next day's OR schedule and disrupting your entire day.

Mark Sandel said...

You know plenty of people fear all of the medical procedures and the entire building too, so it may be a much more complex issue.

Skeptical Scalpel said...

Mark, you could be right. I was actually referring more to physicians' reluctance to order CT scans because of the fear of radiation-induced cancers in the future. I suppose there are some potential patients who share the same concerns.

Post a Comment