COVID-19 and the Loss of Smell: Correlation or Causation?

April 12, 2020 / Jim Fisher, Ph.D.

In a recent Brado Incites post, A.J. Ghergich, Brado’s Chief Technology Officer, shared a synthesis of company research on Google search trends around COVID-19.

The report offered a broad range of analysis and visualization and was published on SEMrush’s influential blog. The full article is available here. It’s sparked interest from the media, healthcare professionals, data analysts and the general public.

Here’s the topline summary: Spikes in symptom-related Google searches strongly correlate to the growth in Coronavirus case counts around the US. The following chart represents this tight statistical relationship:

The findings were based on a focused analysis of search behavior in three states: New York, Michigan and Oklahoma. These three states correspond to different geographical locations where the virus was in various stages of outbreak: peaking, rapid acceleration and early onset.

Also important was Ghergich’s confirmation of Seth Stephens-Davidowitz’s finding that the rise of COVID-19 cases has recently been preceded by a dramatic increase in searches for terms related to a lack of smell or anosmia (ah-naaz-mee-uh). Such searches peaked well in advance of the outbreak’s rapid acceleration.

Ghergich was careful to observe, in reflecting on the significance of the findings, that correlation does not necessarily signal causation. It’s not exactly clear, he said, to what extent Google is “a window or a mirror.”

He offered this exploratory outline of ways researchers could follow up and provide the basis for important advances and applications:

  1. Brado’s research will continue – in the short-run, providing an update of the three-state study with more broad information on all 50 states
  2. Symptom spikes will continue to be monitored
  3. Beyond these preliminary results, Ghergich anticipates additional data analysis could lead the way to:
    • Help allocate or prioritize scarce medical supplies and personnel
    • Anticipate second- or third-wave outbreaks of the virus
    • Refine or re-calibrate current models of the pandemic’s spread

Please follow this blog, as we continue to provide updates. More to come.

Posted in

Jim Fisher, Ph.D.

Jim Fisher, Ph.D., is a strategic advisor at Brado. He is also professor of marketing at Saint Louis University, where he focuses on marketing management, business decision-making and brand communications. He makes extensive use of the case study method in his teaching and consulting, and he has written scores of cases on a wide range of topics, including the healthcare industry, marketing communications and business ethics. He is past president of the North American Case Research Association. He did his graduate education at Yale University and the University of Illinois at Urbana-Champaign.