Journal of Medical Internet Research (Oct 2024)
Association Between School-Related Google Trends Search Volume and Suicides Among Children and Adolescents in Japan During 2016-2020: Retrospective Observational Study With a Time-Series Analysis
Abstract
BackgroundSuicide is the leading cause of death among children and adolescents in Japan. Internet search volume may be useful in detecting suicide risk. However, few studies have shown an association between suicides attempted by children and adolescents and their internet search volume. ObjectiveThis study aimed to examine the relationship between the number of suicides and the volume of school-related internet searches to identify the search terms that could serve as the leading indicators of suicide prevention among children and adolescents. MethodsWe used data on weekly suicides attempted by elementary, middle, and high school students in Japan from 2016 to 2020, provided by the National Police Agency. Internet search volume was weekly data for 20 school-related terms obtained from Google Trends. Granger causality and cross-correlation analysis were performed to estimate the temporal back-and-forth and lag between suicide deaths and search volume for the related terms. ResultsThe search queries “I do not want to go to school” and “study” showed Granger causality with suicide incidences. The cross-correlation analysis showed significant positive correlations in the range of –2 to 2 for “I do not want to go to school” (highest value at time lag 0, r=0.28), and –1 to 2 for “study” (highest value at time lag –1, r=0.18), indicating that the search volume increased as the number of suicides increased. Furthermore, during the COVID-19 pandemic period (January-December 2020), the search trend for “I do not want to go to school,” unlike “study,” was highly associated with suicide frequency. ConclusionsMonitoring the volume of internet searches for “I do not want to go to school” could be useful for the early detection of suicide risk among children and adolescents and for optimizing web-based helpline displays.