**People would rather have the surgical complication to death if either was certain, but they’d rather have a surgery with an increased risk of death the one with the lesser risk of death along with those complications.**

### Exploring how we deal with medical uncertainty

Let’s begin by exploring how we deal with medical uncertainty. Suppose you are a physician trying to communicate with the diabetes patient about different treatment options. In this example and all the others in this article I’ll use hypothetical treatments in hypothetical outcomes just to avoid any specific medical recommendations.

You know that a clinical research trial has shown that a similar medication reduces patient chances of severe cardiovascular disease from 40% to 30% over the next 10 years. How do you give your patient that information? One obvious way to do that is just to relate the scientific study, very factually and slowly.

You could say something like “in randomized clinical trials of patients with similar symptoms life history and demographics taking this medication reduces the overall incidence of cardiovascular disease from 40% in the control group only 30% of the treated group for a period of 10 years following the onset of treatment.

Now I purposefully simplify that it was much less complicated than how the same information would be written in an academic journal or on a medical label, but it was still complicated. You can imagine what the patient might be thinking: “Doctor, I understand that this medication can help some people, but what will it do for me?”

### Do real patients use probabilities when evaluating options

Researchers have examined how real patients use probabilities when evaluating treatment options. When patients are told that the medication has a 40% chance of controlling your disease, some people know what that means: the medication works for 40 people out of every 100, but many people don’t.

They think of that 40% more abstractly as if the physician was relating the chance of rain, it could reflect the physician’s confidence or how much the symptoms would be reduced. Sometimes physicians describe probability’s in verbal labels, saying a side effect is common or rare.

Success of the surgery is likely complications are unlikely. These labels are simpler for people to understand than the probabilities, but different people treat the same label very differently. When people in one study were told that a side effect is very likely and then asked what that means in terms of numerical probability their interpretations range from 30% to 90%.

What do you think a probability label means could have dramatic effects on the treatment you choose? People also cared not just about their own risk but about relative risk how their probability compares to other people’s probability. If a diabetes medication reduces the probability of cardiovascular disease from 40% to 30% it’s not clear how good that is.

### Changes in probability can be misleading

What’s the reference point? For medical decisions people often care about whether their chances are higher or lower than their peers. This can cause problems. Someone who is at the high absolute risk for cardiovascular disease might compare themselves to peers who are similarly overweight, also smoke and also lead a sedentary lifestyle. If we see ourselves as doing better than our peers then we’re less likely to take preventative actions.

Other people’s risk shouldn’t affect our decisions but it does. Even more challenging process is information about changes in risk. Listen to the following hypothetical statement: “Taking a statin medication reduces your long-term risk of cardiovascular disease by 50%”. How should use that information? 50% seems like a big effect!

Does that statement imply that you go from likely to get heart disease too unlikely to get heart disease after you start taking the medication? Probably not. It might mean something much less dramatic. Perhaps you have a 4% chance of heart disease without the medication and a 2% chance with the medication. Emphasizing changes in probability can be very misleading and this problem goes in the other direction too.

Statements like “your chance of heart disease doubles if you don’t take the medication typically cause people to overestimate risks”. What seems to work best is giving people absolute probabilities in presenting those probabilities in terms of frequency. Saying things like “without taking this medication 4/10 people with diabetes like yours will develop heart disease but with the medication drops to only three in 10”.

### Vivid, tangible and defined outcomes influence our decision-making

That is about one in 10 people can prevent heart disease just by taking this medication. Using absolute probabilities makes the consequences of different decisions much more concrete, but it’s not perfect. Remember the common bias in how people use probabilities they overestimate the probability of rare events, underestimate very common events and are largely indifferent for events of intermediate probability.

These biases not only hold when people make decisions about medical outcomes things like changes in life expectancy. They’re even exaggerated. The difference between certainty and a little bit of uncertainty looms very large. In laboratory experiments a treatment that reduces the probability of a disease from 10% to 0% was actually seen as more valuable than another treatment that changes the same disease of probability from 90% to 50%.

We want cures, not changes in probability. Let me emphasize that many medical outcomes are rather rare, even routine anesthesia in healthy adults carries some risk of death but the actual probability is extremely small it’s on the order of one in several hundred thousands. For a healthy adult that’s about the same as a chance of dying in an automobile accident next month.

Not all outcomes are that rare, but many are still very very unlikely. Read the safety sheet for any drug you’ll find a very long list of potential side effects, most of which are a very low probability. These rare outcomes can be vividly brought to mind and we want to avoid them entirely, not have a small chance looming over us. This provides a natural transition to the second factor that shapes medical decisions: the vividness of good and bad outcomes.

**As a general rule outcomes that are more vivid, more tangible and have better defined consequences exert greater influence on our decisions.**