1) You do not have to use numbers or statistics to illustrate risks

DO THIS

A report on influenza quotes a doctor who states that “young children are at higher risk of hospitalization and serious complications of flu.”

DON'T DO THIS

A story about COVID in pregnancy reads, "Infected pregnant women were 50 percent more likely to be admitted to intensive-care units and 70 percent more likely to need a ventilator."

DISCUSSION

Risk is a hard thing to communicate, as individuals do not experience things like “likelihood.” Understanding personal risk based on population-level statistics is therefore very difficult, and in some instances, using numbers to do this is unhelpful. In cases where there is scientific consensus, simply pointing to the existence of a body of scientific literature that demonstrates increased risk is sufficient. A qualifier such as "much more likely" is easier to interpret than a specific percentage.

2) When using numbers to communicate risk, present these as one chance out of a larger number (e.g., “1 out of 20”)

DO THIS

A report summarizes a CDC study by noting that “1 out of 4 older adults who survived COVID” experienced a symptom associated with Long COVID, “compared with 1 out of every 5 people between the ages of 18 and 64.”

DON’T DO THIS

Another report states that “nearly 15 percent of people who’ve contracted COVID-19 say they experienced lingering symptoms.”

DISCUSSION

Audience research indicates that percentages are generally harder to comprehend than probabilities conveyed with a numerator and a denominator. Thus, saying “one out of every six” is better than saying “nearly 15 percent.” While the first example does this fairly effectively, it still works with different denominators, (1 out of 4, versus 1 out of 5). In this case, it would be better still to find a way to use a consistent denominator, as in “4 out of 20” and “5 out of 20.”3) Talk about both absolute and relative risk

DO THIS

A report states that women who frequently use hair straightening products are “twice as likely to develop uterine cancer compared to women who do not use them,” and then compares rates of uterine cancer between these two groups (4.05% for women who frequently use hair chemicals, and 1.64% for those who never use them).

DON’T DO THIS

An article states that women who reported frequent use of hair straightening products “were more than twice as likely to go on to develop uterine cancer compared to those who did not use the products,” without comparing rates of cancer between these two groups.

DISCUSSION

The second example conveys data on relative risk (i.e, the likelihood of cancer in one group compared to others), whereas the first also discusses absolute risk (i.e., the number of people who experience cancer in relation to the entire population). Communicating absolute risk is vital, as without this data, audiences may be confused as to how dangerous a given behavior might be. The phrase “more than twice as likely” suggests that there was a big difference in outcomes between the two groups, whereas in reality, rates of uterine cancer were fairly close in absolute terms.

4) Look for ways to talk about risks on a population-wide scale

DO THIS

A report states that “roughly 1 to 2 percent” of infants with RSV “may need to be hospitalized,” while also observing that the disease causes roughly “58,000 hospitalizations in the United States each year” among children aged 5 or younger.

DON’T DO THIS

A report states that “about one percent of kids with RSV will actually end up admitted to the hospital.”

DISCUSSION

The information conveyed in the second example is accurate, but also slightly misleading, as it creates an impression that the risk of serious illness from RSV is very low. But a small percentage of a big number is a big number. To get this message across, it is helpful to make population-wide extrapolations from statistical data–as the first example does quite effectively.

5) Look for ways to contextualize risks by comparing it to the risks associated with familiar events or behaviors – and not just other health risks

DO THIS

An article on mpox places this in the context of a number of other threats—including the likelihood of being attacked by a shark, being killed in a car accident, and contracting COVID-19.

DON’T DO THIS

A report on Ebola states that the virus causing this disease “is transmissible–but not as transmissible as some other infectious diseases, like Covid-19.”

DISCUSSION

In the media, there is sometimes a tendency to exaggerate the threat certain behaviors or diseases pose—especially if they are new or emerging health threats. Even if these are less dangerous than older or better known threats, we can easily forget this–particularly when the “fear factor” is high. When discussing risk, look for ways to contextualize data–for example, by comparing it to the risks associated with other events or behaviors. Don’t leave it up to audiences to infer magnitude.

By sharing information about various kinds of risks, the first article effectively contextualizes mpox transmissibility rates. By contrast, the second report simply says that Ebola is less transmissible than COVID-19 (which is highly transmissible). This does not really help audiences understand how much of a threat the disease poses.

6) Look for ways to report on the effects of risk reduction measures, and to discuss how effective prevention or treatment options are

DO THIS

A report with the headline “Overdose crisis death toll would be more than double without harm reduction, study says” discusses research aimed at discovering how many lives have been saved through the use of opioid use disorder (OID) treatments like Naloxone (Narcan).

The report highlights that without harm reduction efforts, the death toll from opioid overdoses would have been more than twice as high. It also points out the data that shows “health officials’ response to the overdose crisis in the province is effective.”

DON’T DO THIS

A report on the opioid crisis includes data on prescriptions, overdose deaths, and the likelihood that those addicted to prescription opioids will become addicted to heroin. While it mentions the existence of naloxone (Narcan), and says that this can be used to treat overdoses, it includes no data on how often this drug is used, and says nothing about how effective it is.

DISCUSSION

The tendency to highlight deaths and overdoses is especially present when reporting on stigmatized and/or illegal behaviors. But it is also important to acknowledge effective interventions. Providing statistical data on successes with proven prevention methods (including estimates of how many lives have been saved) can give a more complete picture of the issue, where progress is being made, and promote better health outcomes on both an individual and collective level. Indeed, information that lets people know what they can do to limit risks is associated with positive behavioral change.

7) Draw attention to different kinds of risks – not just those associated with individual characteristics like genetics, lifestyle, or medical history, but also to broader sociocultural and environmental variables.

DO THIS

A report discusses studies that have shown how individuals with a high body mass index (BMI) are at “increased risk for severe COVID-19, including higher rates of hospitalization.” Yet instead of treating this risk as a simple product of individual dietary choice, the report goes on to observe how sociocultural factors are also contributing to the observed outcome. As it notes: “other research shows weight bias can keep larger-bodied people from seeking and receiving appropriate care.” Beyond this, it cites a body of research documenting “weight stigma by healthcare providers.”

DON’T DO THIS

A story reports on the results of a global analysis conducted by a team of researchers at the University of North Carolina, who found that “being obese doubles the risk of hospital treatment from COVID-19, and increases the risk of dying by nearly 50%.” The story goes on to share remarks made by the study’s lead researcher, who “said healthier eating had to be a priority in many countries, with fewer sugary drinks and much less junk and processed food in people’s diets.”

DISCUSSION

The second example treats COVID-19 risk factors as a product of individual choice, forging a link between higher morbidity and mortality rates among obese individuals and the consumption of “sugary drinks” and “junk and processed food.” Yet as social sciences research on this subject has shown, in many cases, the reason obese individuals have experienced worse COVID-19 outcomes owes more to cultural factors (including fatphobia and stigmas associated with being overweight) than personal “lifestyle” factors. As the first example points out, medical professionals often treat obese patients different than those believed to have a “healthy” weight, and this treatment makes overweight individuals less likely to seek out medical assistance for COVID-19 (in addition to many other diseases) until later in the course of their illness than they might otherwise. This is a clear example of a sociocultural risk factor, but failure to report on this feeds into the assumption that obesity in and of itself predisposes one to all manner of diseases (which studies have shown is not the case!). To avoid reproducing assumptions such as these, it’s important to look as broadly as possible at the landscape of health-related risk factors, and to consider sociocultural and environmental variables along with personal ones.

Provide feedback.