JMIR Publications

ABSTRACT

Background:

Previous studies have investigated the association between suicide and the Internet search volumes of terms related to suicide or self-harm. However, their results have varied by age, period, and country, and no studies have exclusively investigated adolescents.

Objective:

This study aimed to determine the association between the Internet search volumes of terms related to suicide/self-harm and the number of suicides among Korean adolescents. Gender differences in this association and the time lag between the Internet search volumes of the terms and connected suicide deaths are investigated.

Methods:

We selected 26 search terms related to suicide and self-harm of Korean adolescents, and the search volumes of these terms for adolescents aged 13–18 years were obtained from the leading Internet search engine in Korea (Naver Datalab). A data set was constructed by combining data from Naver Datalab and the number of the suicide deaths of adolescents on a daily basis from January 1, 2016 to December 31, 2020. Spearman’s rank correlation and multivariate Poisson regression analyses were used to identify the association between the search volumes of the terms and the suicide deaths during that period. The time lag between suicide death and in the increasing trend of the search volumes of related terms was estimated from the cross-correlation coefficients.

Results:

Significant correlations within the search volumes of 26 terms related to suicide/self-harm were observed. The Internet search volumes of several terms were associated with the number of suicide deaths in Korean adolescents, and this association differed by gender. The search volume for “dropout” showed a statistically significant correlation with the number of suicides in all adolescent population groups. The correlation between the Internet search volume for “dropout” and connected suicide deaths was strongest for a time lag of 0 days. In females, “self-harm” and “academic score” showed significant associations with suicide deaths, but “academic score” showed a negative correlation, and the time lags with the strongest correlations were 0 and -11 days, respectively. In the total population, “self-harm” and “suicide method” were associated with the number of suicides, and the time lags with the strongest correlations were +7 and 0 days, respectively.

Conclusions:

The time lag with the strongest correlation between the internet search volume and suicide death was shorter than expected. Therefore, a real-time surveillance system for monitoring changing trends of terms related to suicide/self-harm is suggested with the monitored search terms being gender specific.