"If you take this medication, you will reduce your risk of disease by 20%, and 98% of people taking it have no serious side effects."
Sounds good, right? Perhaps not. What if your chance of getting the disease is only one in 1,000 but two people out of every 100 experience a deadly side effect?
Talk About Risk
Doctors, nurses, reporters and drug advertisers often talk about risk, though they may not call it that. In fact, they often use other terms that sound logical and understandable such as "chance" or "likelihood" (as in "if you smoke, you will increase your chances of having lung disease"). Alternatively, they may couch the terms in personal experience (as in "I think this medicine will help you; it works for most of my patients"). Or, the message may be even more vague: When a doctor says "I think this therapy will help and I doubt it will cause you any problems," the doctor is really saying he or she hopes it helps because it usually does, but that there is a risk that the medication will cause side effects and/or that it will not work.
Whether or not the word is used, these are all examples of risk, the notion that something might or might not happen and that the likelihood is not zero or 100%. I was taught in medical school to "never say never and never say always," because absolute certainty is so rare.
Back to top
Risk Is Not Simple
Risk is a trickier concept than many people think. It may seem like something to avoid at all costs, but the fact is that while you may be able to choose one risk over another, you cannot always avoid risk altogether. Avoiding one risk may pose other risks. For example, if you have high blood pressure and a doctor prescribes a medication to lower it, there is a risk of side effects from the medication; but by avoiding the risk of having a side effect, you increase the risk that you will have a stroke due to untreated high blood pressure.
Many people view any risk at all as a terrible thing, but it is important to realize that it represents a chance that something will happen, not a certainty. Even when talking about bad outcomes (such as the risk of surgery or the risk of medication side effects), there is often an excellent chance that nothing bad will happen.
Consider this example. For a person suffering chronic arthritis and terrible joint pain for the last two years, symptoms probably will continue unless something changes (such as starting an effective treatment). The risk of side effects from any medication is clearly important, but it must be weighed against the high risk that joint pain will continue. If three out of 100 people taking a high-dose anti-inflammatory medication (such as ibuprofen) develop a stomach ulcer each year (that is, if the yearly risk is 3%), it is important to weigh that against the risk of ongoing pain (which is much higher). And it also means that 97 out of 100 people will not suffer an ulcer as they enjoy the benefits of the medication.
Back to top
The Confusing Ways To Describe Risk
The other reason risk is confusing is that there are many ways it can be described. People may choose one way to describe risk to emphasize how high it is, while others use another way to express the same risk to downplay it. Remarkably, the same risk can seem quite different depending on how it is expressed.
One of the most common areas of confusion comes with the difference between absolute risk and relative risk. Imagine that you are offered a choice of two medications to reduce your risk of a heart attack:
- Medication A will decrease your risk of heart attack by 20%.
- Medication B reduces your risk from 5% to 4%.
If you think that Medication A sounds better, you are not alone. Relative risk is described, by comparing the new risk with the risk before treatment. Although vitally important, the actual risk with and without the medicine is not mentioned for Medication A. If you thought Medication B sounds less impressive, perhaps it's because knowing that the starting and ending risks are similar — that is, knowing the actual, or absolute, risks — makes the change seem less dramatic. This is particularly true when overall risk is very low or very high in the first place. Although they provide a more accurate assessment of risk, expressions of absolute risk are often missing from news or ads.
Most of us would think that Medication A is better. But as described, Medication A and B could be the same drug. Reducing a risk of 5 in 100 to 4 in 100 is identical to a 20% relative risk reduction.
So it is not surprising that what you usually read in news reports and see in advertisements are changes in relative risk rather than the more critical number, the absolute risk. The relative risk tends to be the larger and more persuasive number. For the person or advertiser trying to make a dramatic point, most people find it easier to relate to one number (20% risk reduction for medication A) rather than keeping track of the initial risk and how it changes with treatment (5% to 4% for medication B). However, the simplicity of relative risk comes at the cost of incomplete and potentially misleading information. Without knowing where you started, a relative risk reduction of 20% can be nearly meaningless.
Consider one of the most commonly prescribed medications in the world: the birth control pill. For nonsmoking women aged 25 to 29, the risk of death attributed to side effects of the pill is estimated at about 1 in 100,000 and doubles for women who are 30 to 34. If a 33-year-old woman were told she has twice the risk of dying from a birth control pill than a woman under 30, she might incorrectly interpret this as high risk. In fact, even though her risk is slightly higher than for younger women (by 0.001%), she is actually at a very low risk of dying, even lower than the estimated risk of death related to childbirth (10 to 25 per 100,000, also rising with age).
There are, of course, other ways to express risk that shape how it is perceived. Reports of new cases of prostate cancer diagnosed "every three minutes" or the lifetime risk of breast cancer being one in nine, are not very helpful in understanding your own particular risk. Estimating your individual risk by considering factors known to affect it (for example, smoking, which dramatically increases the risk of lung cancer) is a better way to express risk. The generally quoted figure for the risk of getting struck by lightning is 1 in 600,000, but it's much greater if you are holding a metal rod (for instance, a golf club) on a hill during a thunderstorm.
Risk tends to be better understood when broken down by a finite length of time, such as five or 10 years, rather than over one's entire life. Reports of lifetime risk may also include disease that has no bearing on one's overall health, such as a slow-growing prostate tumor that never affects health or longevity (a common scenario among elderly men).
Back to top
The Bottom Line
It is important to acknowledge that risk is easy to misunderstand even with accurate and unbiased information. Cynics say statistics can be used to argue any position, and there is some truth to that. But there are some simple measures you can take to get a better handle on risk:
- Acknowledge that risk is a fact of life and often cannot be avoided entirely. In fact, most decisions one makes about how to prevent or treat disease depend on balancing the risks and benefits of the treatment options.
- Look for reports that put the risk in context. One way is to compare one risk to another, more familiar risk. For example, the risk of contracting a rare disease could be compared with being struck by lightning, while a more common condition could be compared with the risk of injury or death in a car accident.
- Ask yourself if the source is objective and seeks only to inform you, or whether there is another goal in mind. Sometimes it's easy: If the source is an advertiser trying to sell you something, the information is more likely to be biased.
- Beware of reports that use lifetime risk or ones that convert the number of cases per year to cases per day or hour. Even if they are technically accurate, they are probably overemphasizing the point.
What matters most is the absolute rather than the relative risk. In my initial example of the drug with a 20% relative risk reduction with no serious side effects in 98% of users, two fewer people out of every 10,000 would get the disease, but 200 would experience a deadly side effect — clearly an example of a treatment that is worse than the disease. But it's also an example of how misleading descriptions of risk can be.
Back to top
Robert H. Shmerling, M.D. is associate physician at Beth Israel Deaconess Medical Center and associate professor at Harvard Medical School. He has been a practicing rheumatologist for over 20 years at Beth Israel Deaconess Medical Center. He is an active teacher in the Internal Medicine Residency Program, serving as the Robinson Firm Chief. He is also a teacher in the Rheumatology Fellowship Program.