Is Google Trends a useful tool for tracking mental and social distress during a public health emergency?: A time-series analysis

Knipe, D.; Gunnell, D.; Evans, H.; John, A.; Fancourt, D.

Background: Google Trends data are increasingly used by researchers as an indicator of population mental health, but few studies have investigated the validity of this approach. Methods: Relative search volumes (RSV) for the topics depression, anxiety, self-harm, suicide, suicidal ideation, loneliness, and abuse were obtained from Google Trends. We used graphical and time-series approaches to compare daily trends in searches for these topics against population measures of these outcomes recorded using validated scales (PHQ-9; GAD-7; UCLA-3) in a weekly survey (n=~70,000) of the impact COVID-19 on psychological and social experiences in the UK population (12/03/2020 to 21/08/ 2020). Results: Self-reported levels of depression, anxiety, suicidal ideation, self-harm, loneliness and abuse decreased during the period studied. There was no evidence of an association between self-reported anxiety, self-harm, abuse and RSV on Google Trends. Trends in reported depression symptoms and suicidal ideation declined over the study period, whereas Google topic RSV increased (p=0.03 and p=0.04 respectively). There was some evidence that suicidal ideation searches preceded reported self-harm (p=0.05), but graphical evidence suggested this was an inverse association. However, there was statistical and graphical evidence that self-report and Google searches for loneliness (p<0.001) tracked one another. Limitations: No age/sex breakdown of Google Trends data are available. Survey respondents were not representative of the UK population and no pre-pandemic data were available. Conclusion: Google Trends data do not appear to be a useful indicator of changing levels of population mental health during a public health emergency, but may have some value as an indicator of loneliness.