This study provides a comprehensive examination of how lithium, quetiapine, and valproic acid have been represented on X from 2008 to 2022. The findings of the study revealed substantial differences in thematic content, levels of engagement, and types of users across tweets posted in English and Spanish. Moreover, quetiapine was the most frequently mentioned medication; however, lithium generated significantly higher engagement.

Our results align with recent evidence highlighting a growing gap between clinical guidelines and real-world prescribing practices. For example, Singh et al. (2023) published a study involving 10,351 participants across North America, Europe, and Australia, assessing global prescription patterns for BD. The authors concluded that prescribing patterns did not align with major clinical guidelines, revealing notable differences across geographic regions. For instance, there was a higher prescription rate of second-generation antipsychotics and a lower prescription rate of lithium in North America compared to Europe (Singh et al. 2023). These results have also been described in many other regions worldwide (Adiukwu et al. 2025), pinpointing an important concern (Malhi et al. 2020), as lithium is still considered the gold standard treatment for BD in the main clinical guidelines (Yatham et al. 2018; Malhi et al. 2015). These patterns were mirrored in our dataset, where tweets authored by patients predominantly focused on quetiapine, whereas healthcare professionals and academic institutions more often discussed lithium. While social media activity cannot be equated directly with prescribing data, our findings provide complementary insights into the attitudes and perceptions that may influence clinical practice. Previous studies have demonstrated correlations between online discourse and real-world outcomes across different domains. For instance, alcohol-related tweets have been linked to regional consumption rates (Curtis et al. 2018), while social media activity has also been associated with patterns of opioid-related mortality (Sarker et al. 2019) and suicide rates (Wang et al. 2023). These findings suggest that online platforms may capture signals relevant to real-world behaviours. Nevertheless, further research is needed to confirm the consistency and reliability of this complementarity in the context of psychiatric medications.

Our analysis revealed that tweets discussing lithium were not only the most highly engaged but also closely associated with scientific advocacy, particularly in English tweets. These findings may denote a focus of interest and concern within academic and professional communities. However, a high level of trivialisation was also observed in Spanish tweets, in which mocking or sarcastic references were relatively common. Such narratives may foster negative perceptions and reinforce stigma, potentially deterring patients from initiating or continuing lithium treatment. This is especially concerning considering that adherence to medication in BD is low (Coldham et al. 2002; García et al. 2016), and attitudes toward treatment are among the strongest predictors of adherence (García et al. 2016). Our findings support these observations and point out social media as a promising platform for promoting psychoeducation, especially given the high level of engagement regarding tweets talking about lithium, which could help address misperceptions and improve clinician and patient confidence in lithium use.

In addition to lithium, both quetiapine and valproic acid received tweets referencing inappropriate use, with users describing non-therapeutic or recreational consumption. These findings align with literature documenting the misuse of psychiatric medications (Desantis et al. 2008; Schifano et al. 2018; Klein et al. 2017) and highlight the influence of social narratives on public attitudes. This finding suggests careful consideration of the extent to which content related to medications shapes social attitudes and stigma, particularly among patients and their families who may come across such information on social media, which could inadvertently hinder adherence. As previously reported, attitude towards medication is even more crucial for adherence than side effects (Scott and Pope 2002). Furthermore, this result reflects the immense potential of social media to immediately detect such inappropriate use and implement measures as swiftly as possible (Lee et al. 2021).

Interestingly, the proportion of tweets explicitly referring to medication adherence closely mirrored global adherence rates reported in the scientific literature (Coldham et al. 2002; García et al. 2016). This observation highlights the potential of social media analysis to reflect real-world clinical patterns. Previous studies have shown that online conversations can indeed align with epidemiological data; for instance, social media activity related to alcohol and opioid use has been found to correlate with regional patterns of consumption and misuse (Curtis et al. 2018; Sarker et al. 2016). However, further research is needed to confirm the consistency and reliability of these findings across different contexts and populations. Since adherence plays a crucial role in treatment outcomes and is often shaped by individual beliefs and perceptions, increasing the visibility of this issue on platforms like X may be highly beneficial. A greater online presence could foster peer support, improve health literacy, and reduce stigma surrounding non-adherence. Moreover, with the growing implementation of personalised strategies to monitor lithium levels (Zorrilla et al. 2023), strategic dissemination through social media could further support adherence by enhancing awareness and accessibility of these emerging tools.

Another important finding was the prominent discussion over drug shortages of valproic acid among Spanish tweets. The issue of valproic acid availability has already been reported in several Spanish-speaking countries, including Spain and parts of Latin America (CIMA 2023). This valuable information acquired through social media networks is often not as immediate and effective through traditional pharmacovigilance mechanisms (Song et al. 2023), demonstrating the great potential of research in social media networks. The lack of supply is a preventable cause of non-adherence, and proper information about it should be assessed before prescribing a medication as a stabiliser like valproic acid, which is known to be maintained over a long time in BD. Previous studies have mentioned the potential use of social media as a new pharmacovigilance tool that could help, for example, manage the lack of different medications more quickly (Song et al. 2023).

The findings of this study should be interpreted considering several limitations. The analysis was limited to English and Spanish tweets, capturing only a portion of global perspectives and linguistic diversity. Although these languages are widely spoken and provide large, heterogeneous datasets, restricting the analysis in this way may exclude important cultural and linguistic nuances in how lithium, quetiapine, and valproic acid are perceived and discussed. Future studies should extend this approach to additional languages and regions to achieve a more comprehensive understanding of global attitudes toward these medications. In addition, quetiapine and valproic acid are widely prescribed beyond BD. Quetiapine is commonly used across various areas of psychiatry, including schizophrenia, major depressive disorder, anxiety disorders, and insomnia (Taylor et al. 2014). Valproic acid, in addition to its psychiatric uses, is also widely prescribed in neurology for the treatment of epilepsy and other seizure disorders (Tomson et al. 2016). This broad clinical application makes it challenging to determine whether tweets specifically refer to BD or other indications. Therefore, the discourse captured in our analysis should be seen as reflecting the wider social perception of these medications, not solely their use in BD. Additionally, Lamotrigine, although a first-line treatment for bipolar depression, was not included in our analysis because it is not indicated for mania. Future studies should explore the social perception of lamotrigine, considering its importance for the depressive phases of BD, and expand this type of analysis to second-line treatments, including other antipsychotics, to offer a more comprehensive view of the discourse surrounding pharmacological management of BD.

Furthermore, Social Media discourse may overrepresent negative experiences with pharmaceutical treatments (e.g., adverse effects, dissatisfaction), a phenomenon often reported in online settings (Sharma et al. 2020; Leonardo et al. 2020). Our categorisation focused on content types rather than sentiment polarity; therefore, the proportions observed should not be interpreted as a definitive balance of positive versus negative outcomes, nor as a direct reflection of actual prescribing practices. In addition, older adults and socioeconomically marginalised populations are often underrepresented on social media platforms such as X, limiting the generalisability of these findings.

Nonetheless, our study presents several strengths that contribute to the robustness of the findings. First, the analysis covers tweets from 2008 to 2022, providing a comprehensive view of trends and discussions over an extended period. Second, we classified the tweets by user type, allowing for a more nuanced understanding of who engages in these conversations, such as patients or healthcare professionals. Third, by analysing tweets in both English and Spanish, we were able to highlight important linguistic and cultural differences in the discussion of BD medications, offering insights into varying regional perspectives. Finally, the content analysis enabled us to delve deeper into the nature of the discussions, identifying key themes such as medical content and other important areas, such as advocacy or shortages of these medications, which are critical for understanding broader social dynamics surrounding BD treatments.