Danger! Writing about relative risks can lead your readers astray

Published on
May 10, 2010

If you have ever suffered from serious, ongoing pain (RSI, anyone?) you know the desire to take something, anything, to make it go away. What if you were told that you may have a risk as high as 2% of developing heart problems as a result of the painkiller? Would that stop you? And what if you were told that your risk without the drugs was 1%? Would that make you any more likely to start taking the pills?

This would be the scenario were Vioxx still on the market. But it's not. It was pulled in 2004 after a lot of reporting about how it had "doubled the risk of heart attack."

Acting US Food and Drug Administration (FDA) Commissioner Dr. Lester M. Crawford told the Canadian Medical Association Journal at the time that the risk of a patient having a cardiovascular event related to Vioxx was "very small," but that "overall, patients taking the drug chronically face twice the risk of a heart attack compared to patients receiving a placebo."

Partly because clinical trials involving Vioxx were stopped, the true health risk presented by Vioxx has been a bit elusive, prompting one prominent journal editor to say recently that it should still be on the market.

By focusing on "twice the risk," Crawford and so many reporters were giving the relative risk all their attention.  

Relative risk compares different risk levels between two different groups. It can be written as a percent (10%) or a decimal (0.1). Reporting the relative risk is important, but without the proper context – especially the absolute risk – it can inflate the benefits of a drug or device or disguise true threats to public health.

As George Mason University's Stats.org explains:

An important feature of relative risk is that it tells you nothing about the actual risk. This can be very important for evaluating how significant a relative increase might be. A small increase in risk in a large population can result in many deaths. For example, brain tumors are diagnosed in about 6 per 100,000 persons per year, whereas malignant breast cancer is diagnosed in about 134 per 100,000 people. A 10 percent increase (relative risk of 1.1) in brain tumors means .10 x 6 = .6 new cases per 100,000 people.

On the other hand, a 10 percent increase in breast cancer affects 134 per 100,000 people. If the population of the United States is 300 million (which is 3,000 times 100,000), the small increase in brain tumors would result in .6 x 3,000 = 1,800 new cases. In contrast, the same increase of rate in breast cancer would result in 134 x 3,000 = 402,000 new cases, more than 200 times as many.

Relative risk is where drug companies tend to focus their analysis when describing the benefits or downsides of a new product. As Rebecca Goldin points out on Stats.org, the way Merck initially reported the Vioxx data may have made even the relative risk appear much lower than it really was.

A recent study showed that "Women who took aspirin 2-5 times per week had a 23% decreased risk of developing ovarian cancer compared with non-users," according to a press release about the research reprinted on CancerConsultants.com.

A good question about the aspirin study would be, "What was the absolute number of people who developed ovarian cancer without taking aspirin and what was the absolute number of people who took aspirin and did not develop cancer?"

If the risk dropped from a 1% chance of developing ovarian cancer to a 0.77% chance while taking the drug, this would still be a 23% decreased risk.

As a patient, you might want to talk with your physician about other genetic, metabolic or lifestyle factors you may have that increase or decrease your cancer risk before making aspirin a regular habit. Taking aspirin, like any drug, is not risk free.


Tricks of the Trade: Finding Nuggets in the River of Medical Studies