The harms of overtesting

A reprint from a commenter below. This is the best explanation I’ve seen as to why we shouldn’t scan everyone who comes in with abdominal pain. It also explains why not doing a scan and occasionally missing a diagnosis in low-probability cases should not be construed as medical malpractice.

However, will the juries buy such an explanation? Will prosecuting lawyers understand this line of reasoning? Likely not. Thus, the pan-man CT scanning is bound to continue.

(note – emphasis is mine)

The public does not understand that making an eroneous diagnosis is not by itself malpractice.

In this instance, the parents are worried. This is natural, and expected. The degree of parent worry is not in any manner correlated with the likelihood of actual disease. Other markers of parent worry are their presentation to a physician, and demand for further opinions.

All parents whose child is diagnosed with a serious condition ‘knew’ there was something wrong. Unfortunately, many parents whose child proves to have a benign self limited viral condition also ‘knew’ there was something wrong. This is to say that the predictive value of a parent ‘knowing that something was wrong’ is extremely poor. The parents who themselves were wrong forget about it, and certainly do not sue anyone; they may later gripe about the bills that come that suddenly seem completely uncessary because their child has been perfectly healthy ever since.

Physicians treating children see 100 children with belly pain for each one that eventually turn out to have a surgical emergency. The correct test to order depends upon the more likely causes of surgical abdomen. Without ordering CT, ultrasound, barium enema, and other expensive tests on everyone, they use clinical experience. The typical parent has seen no child with a surgical abdomen, while the physician will likely have treated dozens, hundreds or thousands. They will use clinical history, physical examination and a few selected lab studies to assess the likelihood of surgical disease. Often in medicine, if examination at one point in time is insufficient, then repeated exams are warrented, and patients might reasonably be admitted, or allowed to go home with followup the next day.

There is no ego in the process here. There is a reasonable desire not to subject patients to tests with low probability of providing useful information.

Why not order CT for all in this case? The CT capacity at our children’s hospital would have to be more than doubled; negative exams are easy to read so this is easy money for the radiologist and hospital, so we’ll get no argument from them. But, the patient’s third party payers would balk at all these ‘unneccessary’ CT scans; just hear the howling from families who get the whole bill for the denied payment.

There are books on Beyesian decision making. It is far too involved to describe such medical decision making in a simple post here. What it boils down to is that even a good test is of little use if the underlying probability that a patient has the condition being tested for is low. All tests have false positive and false negative results. A good CT for appendicitis will miss 10%, overall 10% and be ‘correct’ the rest of the time. If the random abdominal pain off the street with a 1% chance of appy is subjected to a CT, then the results will be:

0.9% has an appy, confirmed by CT
0.1% has an appy, CT falsely negative
9% no appy, but CT false positive
90% no appy, confirmed by CT

In the end 10 kids have surgery, only one had appendicitis. Nine have operations (with the potential for complications) they did not need.

What happens if we let good clinicians have a crack at the patient first. Nine of ten kids who have surgery actually have appendicitis. This is ok because surgery is not so bad compared with a missed appy, so an extra surgery without appy every now and again is overall more likely to produce the the greatest number of walking talking fully functional human beings in the end. About 25% of appys will have perforated at home, so so much for the informed observant parent. The perforation rate for kids under observation (whether inpatient, or outpatient as in this example) is substantially lower.

So, in the end, why go with an experienced clinician? Because experience does count for something. If you don’t believe it then let me send the intern to your bedside.

Well said.

There seems to be some debate as to the accuracy of CT scans for the diagnosis of appendicitis. A systematic review was done last year with the following results: sensitivity and specificity of abdominal CT scanning for acute appendicitis in adults and adolescents is approximately 94 and 95 percent, respectively.

Thus, the false negative rate is 6 percent; the false positive rate is 5 percent.