Kevin recently posted an article about unnecessary tests, stating "We need to say no to unnecessary tests." I agree with him. As an internist, I am not a procedure guy. My job is to listen to people, make a list of the possibilities, and then make a plan to narrow the list. The two most important things I do when I see a patient are:

  1. Rule out bad things - this is the most important thing. People come because they want to "make sure it is not X." People with enlarged lymph nodes want to know it is not cancer. People with stiff necks often want to know it is not meningitis. People with sore throats want to know if it is strep. You rule things out that would require immediate intervention.
  2. Make a person feel better - Once I have ruled in/out the serious possibilities, I figure out what intervention will make the person better. Sometimes that "intervention" is doing nothing and letting the body do its own thing.

Notice that making the diagnosis is not always the goal. As long as you rule out bad stuff and make the person feel better, finding the diagnosis is gravy.

So the goal of testing is to make a decision: is there a serious problem? What needs to be done to make the person feel better? The goal of testing is not generally to make a diagnosis. If someone has sciatic nerve pain, I order an MRI scan only if there is an intervention needed: so I ask myself, is this person a surgical candidate (the main intervention)?

A few cases I saw recently show that this is not always straightforward.

The first was a man without high risk of colon cancer who asked for a colonoscopy. He had a friend who had it diagnosed at a young age, and was worried about this. Cost was not an issue. So I asked myself:

  • Would a negative test adequately reassure the person that he did not have colon cancer?
  • What is the chance of a false positive test, causing unnecessary worry, and more testing?
  • What is the risk of the test?

A colonoscopy is a good test - it has a very low chance of false negatives (so a negative test is reliable), and the likelihood of a false positive is near zero. In the hands of an experienced gastroenterologist, the risk of the test is fairly low. I ordered the test with a warning about the small risk.

The other case was a woman who was fairly low risk for heart disease who presented with chest pain. My partner saw her earlier in the week and ordered a stress test. The cardiologist actually performed a myoview stress test, which looks at images of blood flow to the heart at rest and with exercise, looking for changes with exercise. The test came back negative.

She came to the office yesterday still having chest pain. She had a few more episodes; one time the pain was worse when she ate a big meal and exerted herself. It went away after resting for five minutes. In this case, I asked myself:

  • With a negative stress test, what is the chance that she still has heart disease?
  • What other diagnoses are possible?

In her case, there was very few other possibilities. I did a lung-function test to rule out asthmatic type symptoms. It was negative. The symptoms were so classic for coronary symptoms, I had to entertain the possibility that the stress test was wrong. There are a few cases when this can happen, either if the blood vessel narrowing is at the back of the heart, or it is very early in the coronary vessels, and so the whole heart was symmetrically decreased in bloodflow. These are causes of false-negative stress tests.

Since this was a significant thing to rule out (potentially a "widow-maker" or left-main coronary), I called the cardiologist and he agreed that the patient needed a catheterization. He called me today and informed me that she had a 95% lesion on the right coronary vessel. He placed a stent.

This demonstrates that stress tests do have a false-negative rate which is not insignificant.  It also showst that the clinical picture should always trump the test in this setting. In this case, the stress test should have been positive, with the lesion. If the catheterization was negative, then the chance of her still having significant heart problems is much smaller, and other possibilities should be pursued.

So for any test, there needs to be a good understanding of:

  • The chance of a false positive
  • The chance of a false negative
  • The pre-test probability of the person having the disease (based on risk factors and the story the patient gives)
  • The risk of the procedure (and the cost).
  • The risk of not making the diagnosis.

Each time I order tests, I think about these things (as should any doctor).

This is why I get frustrated when gnomes from the insurance company say they won't pay for a test. Do they know all of these things on this patient? They just sit in front of a computer and plug the data into a program. They have no idea.

14 Comments