Date Published: February 28, 2019
Publisher: Public Library of Science
Author(s): Yuan Dong, Xin Zhou, Yi Lin, Qichao Pan, Ying Wang, Chao Qiu.
In China, the introduction of antiretroviral treatment has increased the number of people living with HIV/AIDS (PLWHA). New technologies, such as social media, might be useful for enhancing HIV surveillance, especially given the lack of Chinese research, which is related to stigma and discrimination. Thus, the relative anonymity of social media may make it useful for evaluating “hard to reach” PLWHA.
This study used social media data to assess whether it reflected the prevalence of HIV and to explore PLWHA’ needs and online habits.
In 2017, the Baidu Tieba platform was searched to obtain 2,500 HIV-related postings and 2,500 tuberculosis-related postings as a comparative sample. Word clouds and coding schemes were used to analyze the contents and review the users’ needs and online habits. Negative binomial regression was used to evaluate the relationships between word cloud geolocations and provincial numbers of men who have sex with men (MSM) PLWHA cases, after controlling for socioeconomic status.
Word cloud geolocations were associated with reported MSM-PLWHA cases (p<0.001). Over one-third of the HIV-related posts were seeking advice, with 40.12% being related to medical topics, although these posts received the fewest replies. The number of HIV-related social support requests was approximately 3-fold higher than the number of posts providing social support, although relatively similar proportions of support requests and support provision were observed in the tuberculosis-related postings. Social media may help enhance HIV surveillance. Our findings also indicate that the Chinese government, non-government organizations, and healthcare professionals should offer more online support to PLWHA.
According to the Chinese Center for Disease Control and Prevention, approximately 765,000 Chinese people were living with HIV/AIDS (PLWHA) by the end of 2017, which accounted for 0.05% of the total population. Furthermore, five provinces had PLWHA populations that accounted for >0.1% of the total population. The AIDS-related mortality rate in China has been continuously decreasing since the introduction of antiretroviral treatment, which suggests that HIV infection and AIDS are becoming chronic conditions in China, similar to in many developed countries. The increasing longevity of affected individuals are presenting new challenges for health and social services, and a comprehensive management system has been proposed to focus on lifelong treatment and care. However, data regarding PLWHA surveillance, daily habits, and daily needs are relatively limited in China, despite that data being important for guiding a service transition. This may be related to stigma and discrimination making it less likely for PLWHA to participate in surveillance studies, as these factors are prevalent in China. Therefore, innovative approaches are needed to supplement existing tools and improve PLWHA surveillance systems.
Word clouds are visual representations of a body of text, with more frequently used words appearing larger in the word cloud, which can help rapidly summarize textual data in pictorial form. Thus, the present study used word clouds to rapidly summarize the posts, which revealed that geographic words (such as “Chengdu” and “Beijing”) appeared frequently in the HIV-related word cloud (S1A Fig). Furthermore, there were 586 HIV-related posts (25.04%) that included geographic words, relative to only 171 TB-related posts (7.37%) that included geographic words. A map was created to show the posts’ geographical distribution (Fig 1A), which is similar to the findings from a study of Chinese Weibo PLWHA users, who were predominantly homosexual (90.80%). Fig 1B shows a strong positive correlation between the number of province-specific geolocated HIV-related posts and the number of reported MSM-PLWHA cases in the corresponding province (r = 0.84, p<0.001). The multivariate negative binomial regression analysis also revealed a significant positive relationship between number of regional posts per 10,000 web users in each province and the province-level number of MSM-PLWHA cases, with total population and percent with a junior high school education used as the covariates (Table 1). Social media data may be a useful tool for predicting health outcomes[8, 9, 21, 22], and previous studies have used social media data to identify associations between risk behaviors and HIV prevalence. However, to the best of our knowledge, previous studies have only included geolocated data (e.g., geolocated tweets)[8, 21, 22], despite only approximately 1% of tweets being geolocated and only approximately 2% of all users providing access to their geocoded information. Thus, the present study is the first use to unfiltered word cloud data, which extends the literature by exploring whether social media posts are related to disease prevalence. Our findings indicate that the word cloud geolocations in the Baidu HIV-related bar were correlated with the provincial numbers of reported MSM-PLWHA cases (r = 0.84). Interestingly, a lower yet still strong correlation (r = 0.79) was observed between the word cloud geolocations and the total PLWHA population, while a low correlation (r = 0.37) was observed between the word cloud geolocations and the non-MSM PLWHA population. This finding is consistent with previous results, which have indicated that MSM frequently use the internet to seek information[23, 24], and our data indicate that MSM may be more willing to use social media for this purpose than non-MSM PLWHA. Source: http://doi.org/10.1371/journal.pone.0213066