The relationship between internet searches and suicide in Ecuador

PUCE

Team: Diana Mosquera, Francisco Gallegos, Nicolás Acosta

TIME-SERIES ANALYSISGOOGLE TRENDSSUICIDE PREVENTION

Our research, in collaboration with the Faculty of Economics at PUCE, presents an analysis of the relationship between Google search and suicide cases in Ecuador between 2011 and 2021. This study is the first of its kind in Latin America to examine how online search behaviors might correlate with actual incidents of suicide. A total of 131 suicide-related Google search terms were analyzed and grouped into nine categories: general, seeking, prevention, mood, disorders, stress, trauma, minorities, and digital violence. The last two categories -minorities and digital violence- were innovative additions to capture distinct search behaviors that might correlate with suicide. We employed time series analysis to remove autocorrelation and seasonal patterns, avoiding spurious correlations, and then used Pearson correlation and cross-correlation analysis to explore relationships between search trends and suicide rates. To enhance the analysis, we extended our investigation by examining three demographic characteristics: sex, age, and region. This approach revealed unique patterns and relationships specific to different population segments, providing valuable insights for targeted interventions.

The results revealed significant correlations between Google search trends and suicide rates in Ecuador. On average, Pearson correlation coefficients indicated a moderately positive correlation (0.35) between categories, suggesting that suicide cases increase in parallel with the frequency of certain search terms. The category “stress” showed the highest correlation (0.548), indicating that searches related to job uncertainty, financial problems, and relationship problems were closely linked to suicide cases. Other notable correlations were “prevention” (0.438) and “disorders” (0.435). Our cross-correlation analysis with 12 time lags clarified the temporal relationship between search tendencies and suicide cases. Surprisingly, in four of the nine categories, we observed a significant negative cross-correlation at the third lag (-0.191 on average), suggesting a complex relationship in which certain search activities might have delayed protective effects.

Sex analysis revealed that stress-related searches had a much higher correlation with suicide in men (0.554) compared to women (0.149). The category “prevention” showed a substantial correlation in men (0.455) with no significant correlation in women. The age-based analysis highlighted specific vulnerabilities, with the category “mood” showing complex temporal patterns in adolescents and the category “stress” demonstrating significant correlations in young adults. Regional analysis revealed stronger correlations in the highland region. The category “digital violence” showed a distinctive response in women, with a positive correlation at lag 6 and a negative correlation at lag 8, highlighting the gender-specific impacts of online harassment on suicide risk. Similarly, the category “minorities” revealed significant correlations for youth in the highlands region.

This research demonstrates the great potential of digital data to understand and address suicide, a major public health problem. By tracking search trends, health professionals and policy makers can gain valuable insights into collective mental health trends using easily accessible data sources. Negative correlations observed at certain time lags suggest that Internet searches could serve as indicators of risk and, potentially, as elements of prevention. Our findings emphasize the need for tailored suicide prevention strategies that recognize unique risk factors across demographics, particularly addressing the increased correlations observed in young adult man in the highlands region. While Google search data has inherent limitations with respect to people identification, this study represents a significant advance in understanding digital behaviors and suicide risk in Ecuador, establishing a foundation for targeted intervention strategies in diverse contexts, regions, or countries.