Content and trend analysis of user-generated nicotine sickness tweets: A retrospective infoveillance study

Abstract

INTRODUCTION

Exposure to pro-tobacco and electronic nicotine delivery system (ENDS) social media content can lead to overconsumption, increasing the likelihood of nicotine poisoning. This study aims to examine trends and characteristics of nicotine sickness content on Twitter between 2018–2020.

METHODS

Tweets were collected retrospectively from the Twitter Academic Research Application Programming Interface (API) stream filtered for keywords: ‘nic sick’, ‘nicsick’, ‘vape sick’, ‘vapesick’ between 2018–2020. Collected tweets were manually annotated to identify suspected user-generated reports of nicotine sickness and related themes using an inductive coding approach. The Augmented Dickey-Fuller (ADF) test was used to assess stationarity in the monthly variation of the volume of tweets between 2018–2020.

RESULTS

A total of 5651 tweets contained nicotine sickness-related keywords and 18.29% (n=1034) tweets reported one or more suspected nicotine sickness symptoms of varied severity. These tweets were also grouped into five related categories including firsthand and secondhand reports of symptoms, intentional overconsumption of nicotine products, users expressing intention to quit after ‘nic sick’ symptoms, mention of nicotine product type/brand name that they consumed while ‘nic sick’, and users discussing symptoms associated with nicotine withdrawal following cessation attempts. The volume of tweets reporting suspected nicotine sickness appeared to increase throughout the study period, except between February and April 2020. Stationarity in the volume of ‘nicsick’ tweets between 2018–2020 was not statistically significant (ADF= -0.32, p=0.98) indicating a change in the volume of tweets.

CONCLUSIONS

Results point to the need for alternative forms of adverse event surveillance and reporting, to appropriately capture the growing health burden of vaping. Infoveillance approaches on social media platforms can help to assess the volume and characteristics of user-generated content discussing suspected nicotine poisoning, which may not be reported to poison control centers. Increasing volume of user-reported nicotine sickness and intentional overconsumption of nicotine in twitter posts represent a concerning trend associated with ENDS-related adverse events and poisoning.