Using Twitter to Surveil the Opioid Epidemic in North Carolina: An Exploratory Study

Area of Science:

• Public Health
• Epidemiology
• Data Science

Background:

• The opioid overdose crisis has evolved through three distinct phases: prescription opioids, heroin, and synthetic opioids.
• Timely surveillance data is crucial for effective strategies to combat the escalating opioid crisis.
• Social media offers a promising avenue for near real-time public health surveillance.

Purpose of the Study:

• To explore the utility of Twitter data for monitoring the opioid epidemic.
• To assess the correlation between opioid-related tweets and opioid overdose deaths (OODs) in North Carolina.
• To determine if Twitter data reflects the triphasic nature of the opioid crisis.

Main Methods:

• Opioid-related tweets were collected and classified using natural language processing.
• Opioid overdose deaths (OODs) data were obtained from the CDC WONDER database.
• Statistical analyses, including regression and Granger tests, compared tweet patterns with OODs.

Main Results:

• Tweet patterns for heroin and synthetic opioids mirrored the triphasic nature of OODs.
• Tweets mentioning heroin and synthetic opioids were significantly associated with corresponding OODs in the same and following years.
• Twitter data, especially for heroin, showed a predictive relationship with OODs.

Conclusions:

• Twitter data can function as a valuable and timely indicator for opioid overdose mortality.
• The findings support the integration of social media surveillance into public health strategies for the opioid crisis.
• Real-time social media analysis provides crucial insights into evolving drug use patterns and their mortality impact.