Misinformation about an outbreak like Covid-19 is important public health data

One of us (L.H.-D.) heard the first official news of the coronavirus outbreak in China while attending an international conference in Tokyo. Western tourists began donning precautionary face masks and network scientists were brainstorming preliminary analyses. As we all adjusted our travel plans based on the incoming information, it seemed like the virus had reached many residents of Wuhan faster than the news of its emergence.

The spread of information — and misinformation — has been playing a crucial role throughout the unfolding coronavirus outbreak and should serve as a wake-up call for scientists who model epidemics. With every outbreak of a new pathogen comes a race to estimate its transmissibility, which scientists, the media, and the public use to compare the new threat to known enemies.

What this approach fails to capture is the extent to which disease epidemics are shaped by unique interactions between biological and social factors — and that social communication and behaviors during an outbreak are just as important to public health as tests and diagnoses.

Whispers and memes

Covid-19 is a case study in the interplay between contagion, information, misinformation, and behavior. In the early stages of the outbreak, information about how contagious the SARS-CoV-2 virus was censored as “rumors.” Those who shared it attracted the attention of local police, as exemplified by the early warnings from Li Wenliang, the doctor whom the Chinese public consider the whistleblower of the epidemic. Announcements from health authorities, however, drastically underestimated the severity of Covid-19, such as the claim by the World Health Organization in early January that there was “no clear evidence of human-to-human contagion.”

With nothing but whispers to suggest danger, holiday parties and celebrations proceeded apace the week before the Chinese New Year. Most notable was a celebration involving 40,000 families dining together on Jan. 18 in the Baibuting community in Wuhan. Two days later, the severity of the new pathogen was finally announced in a TV news program by Zhong Nanshan, a scientist seen as the leading expert in contagious disease and already well-known for managing the SARS epidemic in 2003.

This information went viral overnight. Citizens and local governments immediately started to take precautions, such as nationally canceling a cultural tradition to visit extended families during the Chinese New Year. Three days later, Wuhan and neighboring cities were under lockdown.

Social media in China has been shaping the course of the disease and the public health intervention. Viral stories of people denied treatment spread meme-like over social media and prompted the government to react by opening temporary hospitals in converted stadiums and convention centers in early February. Within days, these temporary hospitals contained tens of thousands of infected individuals and played a key role in suppressing the outbreak in Wuhan.

We do not yet know how many individuals would have been spared this infection if the virus had been immediately acknowledged and announced. Nor do we know how many more could have suffered had it not been for the viral social media posts and the resulting construction of treatment centers.

As we write this, the parents of one of us (V.C.Y.) have been confined to their home in Wuhan for more than nine weeks. In the initial days of the quarantine, they lived off food they had stockpiled when they first heard whispers of a new virus going around. Later they received groceries from delivery services employees wearing full-body suits and masks.

Like many Wuhan residents, they pass the hours on social media where rumors and misinformation abound despite myth-busting efforts by social media platforms. Some posts tout coronavirus “cures,” usually over-the-counter herbal medications for the common cold, that send people out to medicine shops, standing in line in tight spaces. Conspiracy theories that the virus was genetically engineered, some pointing to the United States as the source, have also been popular.

It is ironic that after Covid-19 became a global problem, the misinformation that fills social media in the U.S. is astoundingly similar.

Science for complex contagions

Although network science provides a framework to study the structure and impact of connections within and across different systems — a way to look at the big picture — it tends to be applied to diseases in a simplistic fashion. Too often it’s assumed that an epidemic must be caused by one pathogen, and stopping that pathogen stops the epidemic.

A recent study co-authored by one of us (L.H.-D.) found that the dynamics of interacting social and biological contagions are not only incredibly different from standard epidemic dynamics but are also much more unpredictable. In fact, when social contagions, like viral news, interact with biological contagions, like viruses, modeling approaches borrowed from the social sciences may be needed to accurately forecast the interacting epidemics. These complex social models can make radically different predictions than classic epidemiological models since they account for critical mass effects, in which a social contagion becomes popular enough for adoption to snowball or where interacting contagions find each other and grow together.

Consider, for example, the measles epidemic in the Philippines that began in February 2019. Over the course of the next 11 months, more than 40,000 people were diagnosed with measles and 500 died. The onset of the epidemic was largely driven by the spread of anti-vaccination sentiment, itself fueled by a dengue vaccine that failed to account for the interplay of dengue strains. This measles outbreak is therefore the result of multiple interacting contagions both within and across pathogens, pushed by biological as well as social forces.

While most disease modelers agree that contagions interact, simply acknowledging this isn’t enough. We need to start collecting better co-infection data, especially during outbreaks such as Covid-19 in which symptoms seem to vary wildly. Direct interactions are possible, as evidenced by the fact that case fatality rate jumps fivefold for people with preexisting respiratory diseases. Could such cases also be more infectious?

Indirect interactions are also to be expected. How many hospital beds are currently occupied because people with the flu are more scared than usual?

We need to start treating misinformation related to an outbreak as public health data in and of itself. This is especially true of misinformation that might affect people’s behavior and put them at risk, such as for-profit efforts to sell ineffective prevention methods. Social media platforms could contribute to disease modeling by publicly sharing data related to the spread of myths and lies around public health emergencies. These data are now often part of the emergency itself and can greatly affect our forecasts and models of interventions.

Until we stop treating epidemics as if they are happening in a vacuum, our models and forecasts will be incomplete. Epidemics occur within an ecosystem of diseases, information, and behaviors, all of which are integral to understanding the spread of Covid-19.

In global health, the big picture always matters.

Laurent Hébert-Dufresne studies contagions and social networks as an assistant professor of computer science at the University of Vermont and a core faculty member of the Vermont Complex Systems Center. Vicky Chuqiao Yang studies mathematical models of human social behavior as a Peters Scholar and Omidyar Fellow at the Santa Fe Institute.