JMIR Publications

ABSTRACT

Background:

Stigma associated with HIV continues to be a major barrier to prevention, testing, treatment, and care despite advances in medical treatments and increased awareness. HIV stigma can influence health behaviors, including reduced likelihood of taking screening tests and decreased adherence to scheduled medical visits. Surveys of the general public in Japan also demonstrated substantial gaps in knowledge of HIV. The content of tweets from the social networking service X (formerly Twitter) has been studied to identify stigmas in mental illness, obesity, and other disorders but has not yet been used to study HIV stigma in Japan.

Objective:

To examine the variations and evolution of stigma related to HIV over an extended period using tweets from X, and to investigate the stigma toward people with HIV associated with various demographic segments.

Methods:

Japanese tweets related to HIV/AIDS posted on X were retrospectively collected; the Phase 1 feasibility study collected tweets from 2011, 2014, and 2017, and the Phase 2 analysis included tweets from each third year from 2011 through 2023. Individual tweets were labeled with the messages they conveyed (stigma and corresponding antistigma types included: labels, marks, responsibility, peril, insults, and fear; tweets without stigma or antistigma messages were considered general education or neutral) along with demographic characteristics and locations; results collected in Phase 1 were used to develop a machine learning model that was applied in Phase 2. The labeled data from Phase 2 were used to answer the study’s research questions concerning yearly changes in HIV stigma and proportions of stigma across population segments.

Results:

A total of 2,016,826 tweets related to HIV/AIDS were identified over the study period; 1,648,556 (81.7%) were from individual accounts, and the remainder were from organizational accounts. In total, 574,687 (28.5%) tweets indicated stigma attitudes, while 1,119,852 (55.5%), 207,320 (10.3%), and 114,967 (5.7%) showed neutral, antistigma, or general education attitudes, respectively. The time series showed the proportion of tweets with stigma (approximately 26% in 2011) increased notably during the COVID-19 pandemic and did not return to the starting level in 2023. Tweets including peril, fear, or insult comprised over 80% of tweets with stigma. The greatest number of tweets were made by people in their 20s, whereas people in their 20s and 60s had the greatest proportions of tweets with stigma.

Conclusions:

This study contributes to the understanding of the types and attributes of HIV stigma expressed in Japan and shows the usefulness of social media for studying stigma. The extent and type of HIV stigma in Japanese tweets changed from before to after the COVID-19 pandemic. Insights from this study can be used to develop future activities and educational programs to combat HIV-related stigma. Clinical Trial: Not applicable.