A couple of weeks ago, Mark Cuban got into an interesting debate with much of the health wonk Twitter community (including me) over whether more lab testing is better. It began when he advocated that everyone get quarterly lab testing. 

While I’m a fan of Cuban’s Shark Tank, and I respect his business acumen immensely, there are a couple things wrong with this. It's worth discussing them in detail. We’re going to do that here today, on Healthcare Triage. 

[intro plays]

First of all, too many people presume that tests are binary things. They're not. When you get a blood test, it doesn't come back "sick" or "well". It comes back with a number value.

Let's say you're looking at white blood cell count, or the number of white blood cells. You might get a reading of seven, which refers to the number of thousands in a milliliter. Is that good? Maybe.

We think that most people should have a value of 4.5 or so, but there's no real meaning to the relative value. Six isn't better than nine, or vice versa.

Let's say you get a 10.5, which is slightly outside the normal range, are you sick? I don't know. White blood cell levels can be affected by so many things, like infections, allergies, or even stress. Do you have symptoms? Do you have other issues? If someone got this value without any other data, I'd have no idea what to do with it - you have to interpret it in context. And that can't be done by a lay-person quarterly.

This is why I teach residents and medical students never, ever to order a blood test unless they are looking for a specific problem. Do you think the patient's anemic? Then it makes sense to check a hematocrit value, which is the number of red blood cells. Are you worried they might have diabetes? Then you should check a glucose value. But getting these tests in a vacuum is rarely helpful at all.

This leads to the second problem. When a lab test picks up something that isn't real, it's called a false positive. It's a lab value that's abnormal, but there really isn't a health issue. When someone is healthy, an abnormal value is much more likely to be a false positive, than a true positive.

This is especially true when a test has a low specificity. And you really should have watched the episodes on sensitivity, specificity, and Bayesian analysis already. They're great. Go ahead, I'll wait.

Specificity is a test characteristic which refers to the proportion of people who are healthy who have a negative test. If it's low, then it means that too many healthy people are having a positive test. A random blood test would have a very low specificity, because an abnormal value would have a disproportionately high probability of being wrong.

This leads to the third problem. What do you do with the abnormal value? As a mentor once said to me, "Ordering a lab test is like picking your nose in public. If you find something, you better know what you're going to do with it."

Most people, even physicians, have a hard time ignoring "abnormal" values. They want to work them up. This leads to excess testing, potential harm, and a lot of money wasted. False positives are just that: false.

You wind up on what a colleague of mine likes to refer to as a diagnostic odyssey, where you chase abnormal value after abnormal value, and never get to the end. I will give you an anecdote to illustrate this.

When I was in medical school, I had some belly pain that was pretty bad. I was even admitted to the hospital, which was overkill, and the pain resolved overnight. I was absolutely fine the next day. But since I was a medical student, they didn't want to miss anything. So I got all these extra tests I didn't need cause I was healthy. But I eventually wound up having every test you could almost imagine, including CAT scans, echocardiograms, x-rays, endoscopy, even cystoscopy, and go look that up to sympathize with me. In each test, something abnormal was found. None of them led to any real problems or diagnoses. They were all pretty much false positives, because I was healthy.

I understand what Cuban's driving at. We've all been told that data are important, and that more is the future. But it's important to understand the difference between using data in the aggregate to generate hypotheses and using data on individuals to make decisions.

Collecting blood from millions of people many times during the year might help us to learn new things about detecting disease in the future; it's great for research. But it's not clear it's good for any one person. We don't know what to do with the results at this point. And if you really think more is better, then why stop at a quarterly lab tests? Why not monthly, or weekly, or even daily?

We should get tests when we have evidence that they will help. We should get them when we expect that they will do more good than harm. At this time, there's no reason to believe that for any individual, quarterly blood tests would fit these criteria.

[outro plays]