What Google searches tell us about interest in the COVID-19 vaccine
Giving policymakers insights on the public reaction to regulatory milestones
Several weeks ago, Google released a compelling ad, “Get back to what you love,” promoting post-pandemic searches and sending a subtle message about the importance of getting vaccinated against COVID-19. Google searches can be a useful near real-time indicator of public interest during a pandemic. We at Surgo have consequently been monitoring Google search trends for insights about how Google can be a beacon for federal and state pandemic response, especially regarding COVID-19 vaccine deployment and uptake.
Our findings have implications that could help federal and state government leaders and public health officials respond more quickly and effectively to people’s needs and worries — whether for this pandemic, or a future one.
Why look at Google Trends?
Researchers have used Google Trends searches as proxies for changes in human behavior during major events. In Part 1 of this series and in a new paper published in JMIR, we recommended Google Trends as a near-real-time monitoring and decision-making tool for policymakers during unpredictable times like this pandemic. For example, we could see where people were still searching for restaurants and bars in the early days of the COVID-19 pandemic, when they really should have been isolating at home:
In this post, the second in a three-part series, we build on this recommendation to show how Google Trends reflects real-time interest in the COVID-19 vaccine in the US. We use Google’s Health Trends API to examine searches for COVID-19 vaccines and vaccine misinformation both across states and over time.
Google searches provide a window to explore COVID-19 vaccine information seeking
Not all COVID searches are alike. Someone could have a general interest in the disease, like daily case rates. Another person could be interested in when the vaccine will be deployed near them. Still another person might be interested in possible side effects of the vaccines.
To begin our analysis, we gathered dozens of possible search terms and grouped them into thematically consistent ‘sentinel indicators’ to track the shifts in search patterns (see our online appendix for full list of search strings). We identified three such indicators:
- ‘Access’ indicator — searches related to availability, timing, and proximity of COVID-19 vaccines
- ‘General’ indicator — searches related to general information, news/updates, vaccine safety, schedules and timelines.
- ‘Risks’ indicator — searches related to potential risks and harm if the vaccine is taken.
We analyzed search patterns as falling into nested categories, per the stylized framework below (segments not to scale).
Finding 1: After overall COVID-19 interest dropped in August, COVID-19 vaccine-specific interest rose after the first Emergency Use Authorization.
Despite a drop in overall COVID-19 search interest, there was a rise in COVID-19 vaccine-specific search interest, especially in the second half of 2020:
This increasing interest is captured by vaccine-specific searches rising from only ~2% to ~43% of overall COVID-19 searches, with noticeable shifts once the first vaccine trial results were announced and EUA was granted.
Finding 2: The rise in vaccine-specific searches was driven by ‘Access’ searches — suggesting people’s strong interest in getting the vaccine after the EUA.
Vaccine-specific searches under the ‘Access’ indicator were driving the increasing trend with searches about when, where, and how to get the vaccine. Surprisingly, the ‘Risks’ indicator searches were relatively negligible, suggesting low search activity related to risk perception and misinformation with the search platform. It is likely that such information seeking is more prominent on social media platforms rather than search engines.
Finding 3: Northeastern states searched for vaccine access at higher rates than other regions.
Both nationally and regionally, access-related searches dominated vaccine information seeking, especially after vaccine trial results and EUA announcements. However, there were marked differences in vaccine search interest patterns between regions. Northeastern states searched for ‘access’ indicators 47% more than the rate of all other regions. This regional gap closed to 31% and 23% when looking at ‘general’ or ‘risks’ searches, respectively.
In our next post, we use further analysis and modeling to unpack these regional differences and connect them to real-word impacts.
BONUS: For the technically-inclined, we also used principal components analysis as an alternative to our handpicked indicators, with relatively unstable results. Hence, we stick with our explicit search clusters.
People were eager to access the vaccine, as measured by their Google searches
In this post on Google Trends during the pandemic, we showed how to group search terms into logical indicators to track public interest in COVID-19 vaccine information. This provides a window into how searches for vaccine access, general information, and risks vary over time, especially after major breakthroughs and regulatory events like release of clinical trial results and EUA. This work builds on previous Surgo Ventures research on Google Trends and provides a useful framework for monitoring public sentiment.
Now, how do these findings connect to real-word phenomena in real-time? Stay tuned for part 3 of this series, in which we statistically model the relationship between the vaccine search interest indicators we developed and real-world vaccine uptake barriers and vaccine rollout speed.
What we learned in this post about Google Trends for vaccines:
- You can use Google Trends to track and monitor the dynamics of specific broad topics around big events within the epidemic/pandemic e.g. vaccines as part of a broader strategy to monitor changes in vaccine interest
- You can use the handpicked sentinel indicators to describe, track, and measure vaccine information-seeking e.g. ‘Access’, ‘General’, and ‘Risks’. These are easy to develop and explain, and they are more stable over time, unlike automatically assigned indicators e.g. PCA components.
- Google Trends is not necessarily the right platform to track some niche topics, e.g. vaccine hesitancy. It is likely that vaccine skeptics are searching for, and receiving, misinformation via social media and private sharing, rather than through search engines like Google.
- Regional variations within this indicator framework are an opportunity to link search interest to potentially meaningful structural/systemic and public health factors
- To do robust analysis on Trends, it greatly helps to have access to the private research API. This gives you an actual percentage of searches for a term, rather than a score normalized between 0 and 100 as you get on the public Trends website.
This is Part 2 of a three-part series. Read Part 1 and Part 3.
This work was made possible by everyone at Surgo Ventures, including but not limited to (in alphabetical order): Daisy Chung, Aaron Dibner-Dunlap, Tich Mangono, and Peter Smittenaar. Additional graphic design support from Katie Armstrong.