Previous: Citronella, Poison Potato Salad, and other Summer Myths
Next: Number Needed to Harm: Treatments Can Hurt You



View count:102,096
Last sync:2024-07-11 04:15
One of the problems with the way we discuss health interventions is that we see them in black and white. Something is either good for you or bad for you. Things are rarely that simple, though. Moreover, there's "good for you" and "GOOD FOR YOU". How do you tell the difference? Watch and learn.

Almost all of the data for this came from the amazing website TheNNT ( You can go there to see those and more.

Additionally, Aaron's new book is out! Please consider buying a copy. He'd really appreciate it!
John Green -- Executive Producer
Stan Muller -- Director, Producer
Aaron Carroll -- Writer
Mark Olsen -- Graphics
One of the problems with the way we discuss health interventions is that we see them in black or white. Something works or it doesn't, something is good for you or it's bad for you. Things are rarely that simple though. Moreover, there's good for you and GOOD for you. How do you know the difference? That's the topic of this weeks' Healthcare Triage.


Let's say your chance of having a heart attack this year is 50%. Let's say I have a new drug that'll reduce that chance to 25%. Since we went for 50% to 25%, I've effectively halved your chance of having a heart attack. That's great, right?

Let's say your chance of developing brain cancer this year is .5%. Let's say I have a new drug that'll reduce that chance to .25%. Again, I've effectively halved your chance of brain cancer. Is that great?

In both cases I've halved your chance of disease. This is known as a relative risk reduction. You take the new risk and divide it by the old risk. In the first instance it's 25 over 50. In the second case it's .25 over .5. In both cases it's one half or 50%. If the drugs are free and have no side effects, then who cares? You should take any kind of risk reduction.

But let's say the drugs are really expensive: then are they worth it? Are those two things the same?

Of course not! In the first case, you had a one in two chance of having a heart attack, those are terrible odds. You absolutely want to avoid that. In the second case, you started with a 1 in 200 chance of having brain cancer. Those are much better odds. You're much more likely to take a chance there, especially if the drugs are expensive or dangerous.

Relative risks are somewhat useless but those are the risks most often reported in news stories or trials. That's because they almost always sound more impressive.

What we really should care about it absolute risk reduction. To calculate that you take the old risk, subtract the new risk, and then divide by 100. Let's work through these two scenarios I've already given you.

With respect to heart attacks we went from 50% to 25%: that's 50 minus 25, or 25, divided by 100. That's .25. So our absolute risk reduction is 25%.

With respect to the brain cancer example, we went from .5% to .25%. That .5 minus .25, or .25, divided by 100. That's .0025 or .25%.

Those numbers aren't even close. The drug for heart attack had an absolute risk reduction of 25%. The drug for brain cancer had a risk reduction of .25%. One is a miracle. The other is much more debatable.

Here's the thing though: almost all the therapies we regard as awesome or necessary have shockingly low absolute risk reductions. They've been sold to you in terms or relative risk reduction but that isn't telling you the whole story.

To the research!

Last you the New England Journal of Medicine published a study touting the positive effects of the Mediterranean diet. Its conclusion was, and I'm quoting: "Among persons at high cardiovascular risk, a Mediterranean diet supplemented with extra-virgin olive oil or nuts reduced the incidence of major cardiovascular events." How much of a benefit was actually seen though? The absolute risk reduction of having a stroke, heart attack, or dying was 1.7%.

A study of high-risk smokers published also in the New England Journal of Medicine in 2011, and I'm quoting again: "Screening with the use of low-dose CT reduces mortality from lung cancer." How much though? The absolute risk reduction was .5%.

What about aspirin to prevent a first heart attack or stroke. No brainer right? I'm sure all of you have heard of that recommendation. The evidence shows though that the absolute risk reduction is .06%.

But let's take absolute risk reduction a step forward. That number can be used to calculate what we call the number needed to treat or NNT. This refers to the number of people we need to give a drug or therapy to in order for one person to receive the benefit.

I know that sounds a little odd. But that's because you've been lead to believe that therapies, like benefits, are black and white. They work or they don't. That's now how the world works. In reality things work on a spectrum. Some people receive a benefit and some people don't. And in the vast majority of cases way more people receive no benefit that people who do.

You calculate a number needed to treat or NNT by taking 100 and dividing it by the absolute risk reduction.

So going back to the heart attack drug which had an absolute risk reduction of 25%: you take 100 and divide it by 25: you get 4. The NNT or number needed to treat is 4. That is, we have to give four people the drug in order for one person to receive the benefit. In this case, a prevented heart attack. That means three of the four people got no benefit at all. None! Two of them would never have had a heart attack and one had a heart attack anyway. Even this miracle drug is three time more likely to give you no benefit than to do what it's supposed to.

The brain cancer drug is much worse! The absolute risk reduction was .25%. So the NNT is 100 divided by .25, or 400. That mean we need to treat 400 people with this drug in order to have one person receive the benefit of a prevented case of brain cancer. 399 out of the 400 who take this drug receive no benefit at all. That's fine if the drug is cheap or it has not side effects. But almost no drugs have those characteristics! So you have to ask yourself: are you OK with being one of the 399? Is a 1 in 400 chance worth it.

The Mediterranean diet, that 1.7% absolute risk reduction translates into an NNT of 61. That means 61 people have to keep to this strict diet for five years for one of them to see a benefit. The other 60 people saw no benefit at all. Is that worth it? I'm not sure. That's for each person to decide. But I bet few people have been told that they're much, much more likely to be doing this for nothing than for something. Lung cancer screening for high-risk smokers with CAT scans has an NNT of 217 to prevent one death. That means that 216 people got scans and all the radiation with no benefit. Worth it?

And you need to treat 1667 people with aspirin for a whole year to prevent ONE first heart attack or stroke. That means 1666 people got treated with the drug for a whole year with no benefit at all. None.

Some of you may think any risk reduction is worth it. Maybe. But you're not considering the harms. That's the topic of NEXT week's Healthcare Triage.