Accepted for/Published in: Journal of Medical Internet Research
Date Submitted: Sep 28, 2023
Date Accepted: Jun 11, 2024
(closed for review but you can still tweet)
Warning: This is an author submission that is not peer-reviewed or edited. Preprints - unless they show as "accepted" - should not be relied on to guide clinical practice or health-related behavior and should not be reported in news media as established information.
Perinatal Polysubstance Use Patterns on Twitter: A Mixed Methods Approach
ABSTRACT
Background:
Approximately 40% of pregnant women with substance use disorders also exhibit polysubstance use (PU) behaviors, yet the associated underlying mechanisms, contexts, and experiences of PU disorder are unclear. Organic information is abundant on social media, such as Twitter, now renamed as X, and traditional quantitative and qualitative methods and natural language processing techniques can be jointly utilized to derive clinical implications.
Objective:
Based on perinatal PU data we extracted on the Twitter (X) platform, we aim to address the following two primary research questions: 1. What is the current trend of PU in perinatal care discussed on Twitter? 2. Are there any distinct patterns in discussion trends of perinatal PU-related tweets before COVID-19, during COVID-19, and during the vaccination period? If so, what are the clinical implications?
Methods:
We utilized the Twitter application programming interface (API) to extract over six million raw tweets from May 1, 2019 to October 31, 2021, containing two or more substance-related keywords provided by our clinical team. In this exploratory study, 7,703 tweets in the United States were randomly selected, and then evaluated using a mixed-method approach. In the quantitative analysis, frequency, trend analysis, and several natural language processing techniques were applied, such as sentiment analysis, to derive statistics to preview the corpus. To further understand clinical insights among these tweets, we conducted an in-depth content analysis with a random sample of 500 PU-related tweets with a satisfying Kappa score for intercoder reliability.
Results:
Both qualitative and quantitative analyses showed that the majority of tweets related to PU contain negative sentiment. There was a larger gap between negative and positive sentiment in the qualitative analysis (negative – 42.11%, positive – 23.98%) versus the quantitative analysis (negative – 48.10%, positive – 42.74%). Through qualitative analysis we were able to identify alcohol and weed as the driving forces among the discovered substance combinations. Additionally, the top three substance combinations were alcohol and drugs, alcohol and weed, and nicotine and weed. The prevalence of tweets pertaining to PU as part of an individual's lifestyle mainly showed support for PU use. Additionally, tweets discussing legal ramification and policies indicated that there is confusion on the legislation involving when a pregnant individual can get charged for a pregnancy resulting in a miscarriage if substance use is involved. Tweets discussing PU had a negative sentiment before COVID-19 and after COVID-19 but before the vaccination period. However, the difference between tweets that were positive and negative was very low, and we found most of the positive sentiment for tweets after the vaccination period.
Conclusions:
While there was a large gap between positive and negative sentiment among perinatal PU tweets in the qualitative analysis, the margin was very small in the quantitative analysis. One important observation is that normalization of PU discussed on Twitter is becoming more prevalent, and thus this study provides some implications on public health policy changes and how individuals with perinatal PU can get proper access to PU interventions and treatments.
Citation
