## Episode Transcript

Bapu JENA: Have you ever been to the grocery store and noticed that the prices of so many different items end in the number nine?

Take for example, mangoes. I love mangoes, so much so that when I go to the store I don’t want to buy just one, I want a whole crate. And if I’m lucky, a crate might cost \$8.99. But it’s not just mangoes. Look around the store. A container of blackberries costs \$3.99. A bunch of green spinach costs \$1.99. A bag of cherries, \$4.99.

There’s a reason that a crate of mangoes is priced at \$8.99 and not \$9. We’re more likely to buy items if they’re priced at \$8.99. And it’s not just because the item is just one cent cheaper.

Say, for example, if a product was \$8.66, you wouldn’t be any less likely to buy it than if it was \$8.67. That’s also just a penny different. But there’s something important about going from \$8.99 to \$9 that makes us feel like one item is cheaper.

It’s because of something that behavioral economists and psychologists have known about for a long time. And it’s called left-digit bias. We all pay more attention to the digit on the left. In this case, do the mangoes cost eight something or nine something?

And it’s not just in the grocery store that left-digit bias is alive and well. Here’s another example. Several years ago, there was an economic study of more than 22 million used car sales. The study’s authors, found two things.

As the number of miles on a used car went up, the price fell, which is no surprise. But what was surprising was that the price took a sharp dip every 10,000 miles. There’s no reason why a car that has 9,900 miles on it should be any different in price than a similar car with 10,100 miles. Yet, they found pretty significant drops in price every time the left-digit changed.

So we see left-digit bias at the grocery store. We see it on used car lots. But what about in the hospital? From the Freakonomics Radio Network, welcome to Freakonomics, M.D.

I’m Bapu Jena. I like to eat fresh mangoes and think about how much they cost. That’s because I’m a medical doctor and an economist. In each episode, I’ll dissect a fascinating question at the sweet spot between health and economics.

Today: What do grocery store prices and heart surgery have in common?

The stakes are much higher in a hospital than in a grocery store. So maybe we wouldn’t expect doctors’ thinking to succumb to something like left-digit bias. But doctors are humans too. And this seems pretty hard wired into how humans think about numbers.

So how much does left-digit bias affect doctors’ decisions? This is the kind of behavioral economics question I nerd out about. And I’m not the only one.

Stephen COUSSENS: I’m trying to remember who was the economist that said something along the lines that economics is essentially painfully elaborating the obvious or something like that.

Bapu JENA: Well, are you — you talking about my wife, or are you talking about somebody else?

COUSSENS: Yeah.

That’s Stephen Coussens. He’s an economist and an assistant professor of Health Policy and Management at Columbia University. A few years back, I saw him present some research at Harvard, where I teach, and he was talking about left-digit bias. It was so fascinating to me. And I literally thought, I wish I had done that. And in fact, I liked it so much that I asked him to work with me on another study that we’ll actually talk about a little bit later. But first, let’s hear about his original study.

COUSSENS:  I had data from a large emergency department in the Boston area. What I had found was, patients showing up in the emergency department just after their 40th birthday were something like 10 percent more likely to be tested for a heart attack than patients showing up just shortly before their 40th birthday.

JENA: There’s nothing biologically different about someone who’s a month before their 40th birthday or a month after their 40th birthday, perhaps unless maybe they went too hard at their 40th birthday party.

So, let me just get this right. When you’re 39 years old and, let’s say, 11 months, and you go to the emergency department for chest pain, you are less likely to get cardiac testing than basically an otherwise similar person who was 40 years old and one month who goes to the emergency department with chest pain. That’s a behavioral economic phenomenon. What’s that called?

COUSSENS: Well, I would say this represents a heuristic. It only makes sense that you would rely on rules of thumb, whether they’re conscious or subconscious, in order to make some of these decisions, like whether to test someone for a heart attack or not. And it seems to be that physicians lump people into bins based on age.

JENA:  When someone’s 39, the mind focuses on the three and, when someone’s 40, the mind focuses on the four. And doctors may categorize patients as being in their quote-unquote 30s versus their quote-unquote 40s, even though you got patients who were basically only different in age by a couple of months.

COUSSENS: Exactly.

After Stephen saw that left-digit bias was happening in one Boston emergency room’s data, he looked at a much larger dataset, and sure enough, he found the bias there too. And that got me wondering, if left-digit bias happens for patients turning 40, what happens when they turn 80? That’s coming up next.

*      *      *

JENA: So, a couple of months after hearing Stephen Coussens talk about his research, I couldn’t shake this question. I was thinking, would we even see even clearer evidence of left-digit bias with older patients? And how might it affect their care? See, the older you are, the more that a few years make a difference. Doctors may view patients in their 70s very differently than patients in their 80s, so much so that a patient who’s just turned 80 years old might be perceived by a doctor to be substantially sicker or higher risk than someone who’s 79 but just about to turn 80, even though these patients are basically the same age.

Take, for example, someone who’s had a heart attack and needs cardiac bypass surgery. That’s a serious surgery. Let me just tell you how serious. When someone comes to the hospital with a heart attack it means that the blood vessels that supply blood to the heart are blocked. And one way that a doctor can treat this is to take a vein from the leg, transplant it to the heart, and bypass that blocked artery. This is called coronary artery bypass graft surgery. It’s a major surgery, and complications can and do happen. Here’s Stephen Coussens again.

COUSSENS: The stakes are even higher in this older population. You would hope this decision, whether or not to offer this surgery to someone, is something that’s being dictated purely by objective probabilities, only the facts. But ultimately, I think there is always a gray area. There’s always this fuzzy subpopulation for which it’s not really clear. And so that’s where I see these heuristics really operating and having the biggest effects.

JENA: So what you and I and our colleagues did was we looked at data from Medicare and we focused on patients who came to the hospital with a heart attack. And we looked at people who are basically 79 years old and 50 weeks versus people who are 80 years old and two weeks, with the thinking that these two groups are basically otherwise pretty similar. And — and we showed that in fact, right? There’s no difference in high blood pressure, cholesterol, diabetes, other cardiac risk factors between these two groups. And so they are, you know, really being exposed to this natural experiment where, just by virtue of having a heart attack in the weeks before or after their birthday, are they going to be subject to left-digit bias on the part of the cardiologist or cardiac surgeon who evaluates them? And what we found is that if you look at people who had a heart attack just after their 80th birthday, those patients were about 25 percent less likely to be recommended cardiac bypass surgery than a patient who had a heart attack just a few weeks earlier. And you might then say, “Okay. Well, how do we know that this is left-digit bias?” Well, we didn’t find any evidence that there was a decrease in cardiac bypass surgery rates when you went from age 76 and 50 weeks to age 77 and two weeks. Similarly, when you go from age 81 and 50 weeks to 82 and two weeks. But we found it at 79 going to 80.

COUSSENS: In this setting that you had just described, it is surprising that this left-digit bias heuristic holds so much sway.

Stephen and I are both economists, and I’m also a doctor; but I’m not a cardiologist. I wanted to bring in the perspective of a working cardiologist to help us better understand what it’s really like in the hospital, treating cardiac patients. Dr. Bryan Smith is an assistant professor of medicine and a cardiologist at the University of Chicago Medical Center.

Bryan SMITH: I clinically manage patients who have advanced heart failure and who we consider for a heart transplant or a mechanical heart pump.

Dr. Smith works directly with cardiac patients. Does he see left-digit bias show up at the hospital?

SMITH: A patient will walk into the E.R. and have a heart attack, and when we’re deciding whether or not to take a patient to the cath lab, sometimes age does play a role. Kidney function plays a role. Hemoglobin, the blood count, plays a role. Left-digit bias affects how we think about all of these things. A creatinine difference between 1.9 and two is not very different, but our brains do think about that differently. A hemoglobin difference of 6.9 versus 7 is not different, but it does affect how we think about these patients.

I love that Dr. Smith is sharing these real-world data points about lab tests like kidney function and hemoglobin levels. These are great examples of the kind of data points that doctors consider all the time when they treat a patient. So mental shortcuts like left-digit bias aren’t just about age-related decisions — like whether to recommend surgery or not — but also the hundreds, and maybe even thousands of decisions that doctors make based on lab tests like these. But Dr. Smith says there are some tools in place to try to help mitigate any one doctor’s biases.

SMITH: So, for example, with the transplant evaluation we do have a lot more time. We have a committee of surgeons, of cardiologists, of nurse practitioners, of pharmacists, social workers, a whole group of people who are at the same table. So, even though some people may be biased in a certain way, we do have a lot of different people from different disciplines who can check us and to make sure that we’re properly assessing these patients and not letting our implicit bias influence the patient’s care.

After Stephen and I found that people who had just turned 80 are less likely to receive cardiac bypass surgery after a heart attack, it raised another important question. How did that affect patients’ outcomes? And here is where it gets really interesting. Looking at data from one year after surgery, we found that there was no difference in the survival rate of patients who got surgery and the ones who didn’t. So, what do we take away from this?

In the real world, when a cardiac surgeon decides not to offer surgery to an 80-year-old who just had a heart attack, there’s usually a reason why. The doctor is usually thinking about other risk factors such as past or current health issues. Left-digit bias simply is not strong enough to make a surgeon not recommend surgery to an otherwise good candidate for surgery. But that bias is strong enough to move the needle on a patient who might be at the border of benefiting from surgery.

If these borderline or marginal patients are the ones who aren’t offered surgery when they have a heart attack after their 80th birthday, that could explain why we don’t see any increase in mortality at age 80, even though fewer patients get surgery. It’s because the surgeries that don’t get performed are for people who would’ve been unlikely to benefit anyway. I asked Stephen: After seeing the results of our study, what would he say to a cardiac surgeon with a patient who is 79 or 80 years old?

COUSSENS: In one respect, I think this is somewhat intractable, because I think it is very much just a natural part of the way people think. But, that said, I think there are ways to soften the sharpness of these physician behaviors. So, using additional decision support tools. There’s also just the simple prompting of, well, would you feel differently if this patient were 79 and not 80, or if this patient were 40 and not 39? And if the answer is yes, then perhaps reconsider your course of action.

JENA: Yeah. You know, there’s this idea of “cognitive de-biasing.” It’s the idea that if you educate people about the biases that they make, they may be subsequently less likely to make them. So, you know, you can imagine in this context that if surgeons were made aware of this left-digit bias in decision-making, that they may be less likely to do it in the future. Now, I should say that in the context of what we found, I don’t know that that would be a good thing, right? If we eliminated left-digit bias in cardiac bypass surgery, and instead what happened was that patients who were 80 years old were no longer 25 percent less likely to get cardiac bypass surgery than patients who were 79 years old, we basically just increased the number of 80-year-olds with heart attacks who are getting cardiac bypass surgery. And if our study is correct, we wouldn’t have any improvements in survival. So that, so that wouldn’t be a good thing.

COUSSENS: I totally agree. It’s not clear that this bias is always bad, or that it’s something that we can fundamentally alter. But through research, if we can identify the specific scenarios in which this creates a problem, we can intervene in those scenarios.

We started off today talking about the role of a specific behavioral bias in how doctors think. That’s left-digit bias. There are, of course, many other behavioral biases that we could have talked about, but I wanted to end today’s show by recognizing that any discussion of bias in medicine would really be incomplete if we didn’t recognize the broader impact that structural biases with respect to race and gender have on the care that our patients receive.

We’ll talk more about that soon, I promise. But for now, that’s it for Freakonomics, M.D. Thanks for listening, and I hope you subscribe to or follow the show. If you have thoughts on the show, ideas, anything at all, I’d love to hear from you. You can email me at bapu@freakonomics.com. That’s B A P U at freakonomics dot com.

*      *      *

Freakonomics, M.D. is part of the Freakonomics Radio Network, which also includes Freakonomics Radio, No Stupid Questions, and People I (Mostly) Admire. This show is produced by Stitcher and Renbud Radio. You can find us on Twitter and Instagram at @drbapupod. This episode was produced by Colleen Pellissier and mixed by Adam Yoffe. Our staff also includes Stephen Dubner, Alison Craiglow, Greg Rippin, Joel Meyer, Tricia Bobeda, Emma Tyrrell, Lyric Bowditch, Jasmin Klinger and Jacob Clemente. If you like this show or any other show in the Freakonomics Radio Network, please recommend it to your family and friends. That’s the best way to support the podcasts you love. As always, thanks for listening.