Tracking COVID-19 using taste and smell loss Google searches is not a reliable strategy

Kim Asseo, Fabrizio Fierro, Yuli Slavutsky, Johannes Frasnelli, Masha Y. NivScientific 2020 · DOI

Abstract

Web search tools are widely used by the general public to obtain health-related information, and analysis of search data is often suggested for public health monitoring. We analyzed the popularity of searches related to smell loss and taste loss, recently listed as symptoms of COVID-19. Searches on sight loss and hearing loss, which are not considered COVID-19 symptoms, were used as controls. Google Trends results per region in Italy or state in the US were compared to COVID-19 incidence in the corresponding geographical areas. The COVID-19 incidence did not correlate with searches for non-symptoms, but in some weeks had high correlation with taste and smell loss searches, which also correlated with each other. Correlation of the sensory symptoms with new COVID-19 cases for each country as a whole was high at some time points, but decreased (Italy) or dramatically fluctuated over time (US). Smell loss searches correlated with the incidence of media reports in the US. Our results show that popularity of symptom searches is not reliable for pandemic monitoring. Awareness of this limitation is important during the COVID-19 pandemic, which continues to spread and to exhibit new clinical manifestations, and for potential future health threats.

TL;DR

While Google Trends for COVID-19 symptoms showed an initial high correlation with infection rates, this relationship "decouples" over time due to media influence and public saturation. Consequently, relying on simple search volume for pandemic monitoring is an unreliable strategy that risks missing significant surges in cases.

What Problem Does This Paper Address?

Early in the COVID-19 pandemic, laboratory testing was scarce and slow. Public health researchers looked toward "digital epidemiology"—using internet search data to monitor disease spread in real-time (similar to the famous, though eventually flawed, Google Flu Trends).

Because loss of smell (anosmia) and taste (ageusia) emerged as "signature" symptoms of SARS-CoV-2, several groups hypothesized that spikes in searches for these symptoms could predict new hotspots. The authors of this paper set out to test the robustness of this hypothesis: Does the correlation hold up across different countries and, more importantly, does it persist over time as the pandemic evolves?

Methods

The researchers compared Google Trends data for symptom-related searches against official COVID-19 incidence rates in Italy (by region) and the US (by state).

  • Keywords: They tracked "taste loss" and "smell loss" (and their Italian equivalents).
  • Controls: To ensure they weren't just seeing a general increase in health-related searches, they used "sight loss" and "hearing loss" as controls.
  • Media Analysis: They used Media Cloud, a platform that tracks digital news, to quantify how often these symptoms were mentioned in the press.
  • Statistical Analysis: The team employed Pearson correlations and a sliding window analysis (31-day windows shifted by one day) to see how the relationship between searches and cases changed throughout the first half of 2020.

Key Insights and Ideas

  • The "Media Effect" is a Major Confounder: In the US, search volume for smell loss was more strongly correlated with news reports than with actual new cases. People weren't just searching because they had the symptom; they were searching because they read about it.
Comparison of COVID-19 cases, media reports, and Google search trends for taste and smell loss in Italy and the US.
  • Initial Success was a Fluke of Timing: In the very early stages (March in Italy, April in the US), correlations were incredibly high (up to 0.97). However, this was the period when the symptom was first being "discovered" by the public.
  • Decoupling During the Second Wave: The most damning evidence against the strategy appeared during the US "second surge" in June/July 2020. While cases skyrocketed, search interest for smell and taste loss remained relatively flat or fluctuated wildly, completely failing to reflect the actual epidemiological reality.

Key Results

  • Correlation Breakdown: In the US, the correlation between smell loss searches and new cases was high in early April but dropped significantly and fluctuated between positive and negative values by June and July.
Sliding window correlation between new COVID-19 cases and search popularity, showing significant decoupling in the US (purple boxes).
  • Quantitative Media Link: In the US, the pair-wise correlation between media mentions and search volume was 0.50 for smell loss (p < 0.001), significantly higher than the correlation found in Italy (0.17).
  • Control Stability: Searches for sight and hearing loss showed consistently low or non-existent correlation with COVID-19 cases, proving that the smell/taste spikes were specific to the disease interest, even if they weren't accurate for tracking.

Limitations and Future Work

The authors acknowledge that while simple search volume is unreliable, "sophisticated" models that account for media saturation and "noise" might still have value. However, they warn that as a symptom becomes common knowledge, the "novelty" wears off, and people stop searching for it even if they experience it.

Future Directions: * Moving away from self-reporting tools toward independent biological markers, such as sewage monitoring (wastewater epidemiology). * Developing better algorithms to filter out "curiosity searches" from "symptomatic searches" in digital data.

Summary updated

Abstract

Web search tools are widely used by the general public to obtain health-related information, and analysis of search data is often suggested for public health monitoring. We analyzed popularity of searches related to smell loss and taste loss, recently listed as symptoms of COVID-19. Searches on sight loss and hearing loss, which are not considered as COVID-19 symptoms, were used as control. Google Trends results per region in Italy or state in the US were compared to COVID-19 incidence in the corresponding geographical areas. The COVID-19 incidence did not correlate with searches for non-symptoms, but in some weeks had high correlation with taste and smell loss searches, which also correlated with each other. Correlation of the sensory symptoms with new COVID-19 cases for each country as a whole was high at some time points, but decreased (Italy) or dramatically fluctuated over time (US). Smell loss searches correlated with the incidence of media reports in the US. Our results show that popularity of symptom searches is not reliable for pandemic monitoring. Awareness of this limitation is important during the COVID-19 pandemic, which continues to spread and to exhibit new clinical manifestations, and for potential future health threats.
BibTeX
@inproceedings{asseo2020tracking,
  author = {Kim Asseo and Fabrizio Fierro and Yuli Slavutsky and Johannes Frasnelli and Masha Y. Niv},
  title = {{Tracking COVID-19 using taste and smell loss Google searches is not a reliable strategy}},
  booktitle = {Scientific},
  year = {2020},
  doi = {10.1038/s41598-020-77316-3},
}
Figure