The Case for Tracking Misinformation the Way We Track Disease

"The world is better prepared to address vaccine misinformation than ever before. Now public health must be held just as accountable as the social media companies for taking action."

Misinformation threatens the public's health, the credibility of institutions, and the safety of experts and frontline workers. However, few public health organisations are capable of tracking and responding to disease-related misinformation as it happens. This commentary explores how big data - specifically, publicly available media data - can facilitate efforts to understand and respond to misinformation. Since 2017, The Public Good Projects (PGP) has used supervised machine learning to aggregate and code millions of conversations relating to vaccines and the COVID-19 pandemic broadly, in real time. The authors examine PGP's experiences with, and learnings from, this monitoring.

An example of PGP's work is Project Vaccine Communication Tracking and Response (VCTR), a full-time effort to quantify and categorise all communications related to vaccines across a broad range of media. Public health researchers supervise this process daily and provide communications recommendations to practitioners across a range of disciplines, including health departments and health systems, non-governmental organisations (NGOs), researchers, and journalists.

Through this and other types of monitoring, PGP has observed that, since 2020, vaccine opponents have become emboldened, capitalising on hesitancy regarding novel COVID-19 vaccines to sow distrust in all immunisations, as well as in health authorities, government, and pharmaceutical companies. However, PGP analyses demonstrate that "the sources of vaccine misinformation are readily identifiable. Dominant talking points originate from a handful of highly influential, well-connected accounts....The narratives found within vaccine opposition have remained fairly consistent when compared to time periods prior to the pandemic, with one notable exception being the targeting of health authorities..."

Just as vaccine misinformation can be tracked, so can attacks on health authorities. PGP analysis indicates a 217.8% increase in mentions of national health authorities by vaccine opponents on Twitter, compared to the nine months preceding March 2020 (when COVID-19 lockdowns began in many countries). These vaccine opponents have paired their anti-vaccine messages with criticism of individual health experts, which seems likely to have contributed to the growing attacks, including physical threats, against them. The authors worry that "The societal impact of this concerted effort to sow distrust in health authorities and institutions will likely have long-term consequences."

Public health departments are ill equipped to counter misinformation, especially while fighting the COVID-19 pandemic, due to reasons such as the fact that communication is secondary to the technical research and fieldwork unique to public health. Thus, the authors argue, cross-sector collaboration at the individual, organisational, and policy levels will be crucial to the success of the field of "infodemiology". For instance, they suggest that crisis communications' frameworks and tools from other disciplines could add value to how public health tackles misinformation. While sector and organisational nuances exist, crisis communication workflows generally fall into three main phases: pre-crisis planning, crisis response, and post-crisis learning. Media monitoring and big data analysis can be a resource within each phase, as outlined here.

An example cited here that is freely accessible is the "Vaccine Misinformation Management Field Guide" (see Related Summaries below). Created through a partnership between the United Nations Children's Fund (UNICEF), First Draft, the Yale Institute for Global Health, and PGP, this tool provides guidance on developing strategic national action plans informed by social listening to counter misinformation about vaccines globally.

PGP draws on this work in offering recommendations for addressing misinformation, such as:

• Sources of vaccine misinformation are known; there is a need to operationalise learnings and engage the pro-vaccination majority in debunking vaccine-related misinformation.
• Existing systems can identify and track threats against health experts and institutions; health departments and public health programmes should consider monitoring of threats against their health workers with the same regularity and normalcy as they have for monitoring disease.
• Universities with public health or health communications programmes should incorporate infodemiology as a core aspect of coursework, and current public health workers should be trained to analyse health-related media data in a way that becomes as routine as their monitoring of disease-related data.
• Responses to misinformation should draw from cross-sector crisis management best practices and address coordination gaps.

In conclusion: "Real-time monitoring and addressing misinformation should be a core function of public health, and public health should be a core use case for data scientists developing monitoring tools. The tools to accomplish these tasks are available; it remains up to us to prioritize them."