A Dataset of Patients Reviews on Doctissimo Forum of the 2017 Levothyrox Scandal in France for the Early Detection of Safety Signals from Social Networks

In France, the Levothyrox case is an extremely mediated pharmacovogilance case which received a lot of media coverage in August 2017 following the change, in March of the same year, of the pharmaceutical formulation of the specialty (modification of the type and content of excipients at the initial request of the french's national agency of drug safety - ANSM). Many patients have experienced adverse effects related to fluctuating thyroid function when they were already treated and stabilized with Levothyrox (3 million french patients with thyroid pathologies). This dataset presents patients posts collected from the French health forum Doctissimo, Thyroid and Endocrine Problems sub-forum using a web scraping algorithm. The main idea is to find out whether the result of the processing of the data exchanged on the Doctissimo forum can be used in pharmacovigilance and whether it would have allowed better reactivity from the public authorities and responsible laboratories. <br> Doctissimo was chosen since it is the most used health forum in France by drug consuming patients (ranking first with 61\% of users). Other sources of information could have been chosen such as Twitter which is the most used website in the world by drug consuming patients. Twitter brings together 52% of these patients against 27% for all discussion forums combined. However, access to Twitter data is chargeable, which is why Doctissimo has been selected. The extraction was performed on the ``Thyroide et Problemes Endocriniens (Thyroid and Endocrine Problems)'' sub-forum with the keyword ``levothyrox''. The choice of extracting information from this forum using the particular relevant keyword ``levothyrox'' was to limit the amount of extracted data. Indeed, during data extraction, Doctissimo blocks the scraping task when reaching a limit of 8,000 extracted discussion threads, since it detects an automatic machine activity. We collected the messages written between years 2000 and 2020. This resulted in a total of 110,260 comments written by a total of 7650 subjects. For each of the comments, we extract the date, pseudo of the person who wrote the comment, the comment's text, and URL link.法国左甲状腺素（Levothyrox）事件是一起广受媒体关注的典型药物警戒（Pharmacovigilance）事件。2017年3月，应法国国家药品安全署（ANSM）的初始要求，该药品的处方制剂进行了调整（变更了辅料的种类与含量），同年8月该事件引发大量媒体报道。当时已接受左甲状腺素治疗且病情稳定的法国患者（全法约300万甲状腺疾病患者）中，多人出现与甲状腺功能波动相关的不良反应。本数据集收录了通过网络爬虫算法从法国健康论坛Doctissimo的「Thyroide et Problemes Endocriniens（甲状腺与内分泌疾病）」子版块抓取的患者发帖内容，旨在探究：对Doctissimo论坛上的交流数据进行处理后得到的结果，是否可应用于药物警戒领域，以及该结果是否能够帮助公共主管部门与相关制药企业提升响应效率。 之所以选择Doctissimo论坛，是因为它是法国服药患者最常使用的健康论坛，用户占比达61%，位居行业第一。理论上也可选择其他信息源，例如Twitter——它是全球服药患者使用最多的网站，覆盖52%的该类患者，而所有综合讨论论坛的总覆盖占比仅为27%。但Twitter的数据获取需付费，因此最终选择了Doctissimo论坛。本次数据抓取针对的是Doctissimo论坛的「Thyroide et Problemes Endocriniens（甲状腺与内分泌疾病）」子版块，抓取关键词为「levothyrox」，选择该相关关键词进行抓取的目的是控制提取数据的总量。事实上，在数据抓取过程中，当抓取的讨论帖数量达到8000条阈值时，Doctissimo会检测到自动化机器行为并阻断抓取任务。本次抓取的发帖时间范围为2000年至2020年，最终共收录7650名用户发布的110260条评论。每条评论均提取了发布日期、发布者昵称、评论正文以及URL链接。