A Twitter account corrects one of the biggest problems in health reporting

Not all risk is reported equally.
Not all risk is reported equally.
Image: AP Photo/David Duprey
We may earn a commission from links on this page.

In health news, one of the key bits of information to communicate is risk, or the chances that, given one difference or intervention, a particular outcome will occur.

The trouble is, the way many press releases and media outlets report on risk can be highly misleading. By giving readers only the relative risk of a specific outcome, they tend to drastically overstate the likelihood of the studied outcome.

Fortunately, an epidemiologist studying in Australia has come up with a simple way to clarify risk communication: A twitter account called @justsayrisks. Similar to the highly successful @justsayinmice—a Twitter account created by computational biologist James Heathers to illustrate how often the progress of studies conducted on rodents is overstated in the media—@relativelyrisky retweets news stories based on studies with both the relative and absolute differences in risk.

The account is the creation of Gideon Meyerowitz-Katz, who is currently pursuing his doctorate at the University of Wollongong in Sydney, Australia. Meyerowitz-Katz keeps a Medium blog where he routinely clarifies misleading health news stories by explaining how the science was actually conducted. With @relativelyrisky, which he created in April 2019, he has an opportunity to explain how risks are actually calculated.

Risk assessment comes from two kinds of studies: clinical trials or cohorts. Clinical trials compare what happens to a group of human participants who receive a new medicine or procedure compared to a group that receives a placebo or current standard treatment. Cohort studies track a group of people over time, and account for self-reported differences in behavior, like smoking status, diet, or exercise regime. Based on what happens to people in each group with each intervention, researchers can give a statement on increased or decreased likelihood of an outcome, like developing heart disease or dying.

“Relative risk is the ratio between one risk and another. Basically this means that you take the likelihood of one event happening and divide it by another,” he wrote last week. Any relative risk over 1 means that one group was more likely to have some health outcome occur. If the relative risk of having shredded furniture for cat owners was 1.50 compared to dog owners, you could say that cat owners were 50% more likely to have shredded furniture than dog owners. (Relative risk than 1 means that an outcome is less likely to happen.)

Absolute risks are often a lot smaller than relative risks, because they represent the difference between the percentages of a given outcome in two groups. So, if 10 out of 100 cat owners have shredded furniture, but only five out of 100 dog owners do, cat owners have an absolute risk of 5% of having ruined furniture compared to dog owners.

Understanding the difference in these risks in health reporting is critical, because often in cohort studies, scientists are looking at understanding overall risk of death or developing cancer. As Meyerowitz-Katz points out, for most of us, these absolute risks are low. Yet if a study shows that some people have a high relative risk of dying or developing cancer based on a particular choice, like eating red meat, it makes red meat consumption seem incredibly dangerous. Absolute risks show that is some danger to eating red meat, but account for background rates of cancer.

Risks are a tricky concept for anyone. It’s often useful to use some other baseline as a comparison, which is what relative risks aim to do. However, it’s important to remember that everything comes with risks. As David Spiegelhalter, a statistician at the University of Cambridge, has pointed out, we all have a 100% risk of death by virtue of being alive.

Update: The Twitter handle referenced in the article has been updated from @relativelyrisky. It is now @justsayrisks.