Google Trends™ and Quality of Information Analyses of Google™ Searches Pertaining to Concussion

Abstract

Sports-related concussions occur with high incidence in the United States. Google Trends™ (GT) analyses indicate changes of public interest in a topic over time, and can be correlated with incidence of health events such as concussion. Internet searches represent a primary means of patient education for many health topics, including concussion; however, the quality of medical information yielded by internet searches is variable and frequently of an inappropriate reading level. This study therefore aims to describe GT over time and evaluate the quality and readability of information produced by Google™ searches of the term "concussion." We identified a strong negative correlation from 2009 to 2016 between GT scores and total number of American high school football participants (R ² = 0.8553) and participants per school (R ² = 0.9533). Between 2004 and 2020, the monthly GT popularity score were variable (p = 3.193E-08), with September having the greatest scores, correlating with the height of American tackle football season. Applying five validated quality assessment scoring systems at two time points, it was confirmed that different sources yielded varying quality of information. Academic and non-profit healthcare sources demonstrated the highest quality metrics across two time points. There was significant variability of scores among the different scoring systems, however. The majority of searches at both time points yielded information that was rated as "fair" to "poor" in quality. Applying six readability tests, we revealed that only a single commercial website offered information written at or below the American Medical Association- recommended 6th-grade level for healthcare information. In summary, GT data analyses suggest that searches correlate with the American tackle football season and increased between 2009 and 2016, given that public interest in concussion increased and annual participation in football decreased. The quality of information yielded by Google™ searches and readability are inadequate, indicating the need for significant improvement.

Keywords: Google Trends™; concussion; online searches; quality of information; readability.

FIG. 1.

Association of concussion Google Trends™ scores with the U.S. 2009–2016 annual high school participation in all sports and football. Pearson's correlation coefficient was calculated in Microsoft Excel (Microsoft Corporation, Redmond, CA) between the average yearly Google Trends™ scores for “concussion” and yearly participation data derived from the National Federation of State High School Associations. Results show that football participation has a strong negative correlation with Google Trends™ scores whereas all sports has a strong positive correlation.

FIG. 2.

Association of concussion Google Trends™ scores with the U.S. 2017–2019 annual high school participation in all sports and football. Pearson's correlation coefficient was calculated in Microsoft Excel (Microsoft Corporation, Redmond, CA) between the average yearly Google Trends™ scores for “concussion” and yearly participation data derived from the National Federation of State High School Associations. Results show that football participation has a strong negative correlation with Google Trends™ scores.

FIG. 3.

Month-specific statistically significant seasonality of Google Trends™ data. Monthly variability calculated with one-way ANOVA regression analysis from 2004 to 2021 with α = 0.05 using Microsoft Excel (Microsoft Corporation, Redmond, WA). January, March, June, August, and September had statistically significant seasonality over the time period analyzed, suggesting that certain months had seasonally increased or decreased search volumes. ANOVA, analysis of variance.

FIG. 4.

Month-specific z-scores of normalized concussion Google Trends™ scores from January 2004 to May 2021. Average monthly z-scores were calculated in Microsoft Excel (Microsoft Corporation, Redmond, CA). Fall and winter months, September through February, showed above-average z-scores whereas the spring and summer months, March through August, showed below-average z-scores, showing that there is more public interest in concussion in the fall and winter months.