Exploring the vaccine conversation on TikTok in Italy: beyond classic vaccine stances.

TikTok, a social media platform for creating and sharing short videos, has seen a surge in popularity during the COVID-19 pandemic. To analyse the Italian vaccine conversation on TikTok, we downloaded a sample of videos with a high play count (Top Videos), identified through an unofficial Application Programming Interface (consistent with TikTok's Terms of Service), and collected public videos from vaccine sceptic users through snowball sampling (Vaccine Sceptics' videos). The videos were analysed using qualitative and quantitative methods, in terms of vaccine stance, tone of voice, topic, conformity with TikTok style, and other characteristics. The final datasets consisted of 754 Top Videos (by 510 single users) plus 180 Vaccine Sceptics' videos (by 29 single users), posted between January 2020 and March 2021. In 40.5% of the Top Videos the stance was promotional, 33.9% were indefinite-ironic, 11.3% were neutral, 9.7% were discouraging, and 3.1% were ambiguous (i.e. expressing an ambivalent stance towards vaccines); 43% of promotional videos were from healthcare professionals. More than 95% of the Vaccine Sceptic videos were discouraging. Multiple correspondence analysis showed that, compared to other stances, promotional videos were more frequently created by healthcare professionals and by females, and their most frequent topic was herd immunity. Discouraging videos were associated with a polemical tone of voice and their topics were conspiracy and freedom of choice. Our analysis shows that Italian vaccine-sceptic users on TikTok are limited in number and vocality, and the large proportion of videos with an indefinite-ironic stance might imply that the incidence of affective polarisation could be lower on TikTok, compared to other social media, in the Italian context. Safety is the most frequent concern of users, and we recorded an interesting presence of healthcare professionals among the creators. TikTok should be considered as a medium for vaccine communication and for vaccine promotion campaigns.

Understanding the vaccine stance of Italian tweets and addressing language changes through the COVID-19 pandemic: Development and validation of a machine learning model.

Social media is increasingly being used to express opinions and attitudes toward vaccines. The vaccine stance of social media posts can be classified in almost real-time using machine learning. We describe the use of a Transformer-based machine learning model for analyzing vaccine stance of Italian tweets, and demonstrate the need to address changes over time in vaccine-related language, through periodic model retraining. Vaccine-related tweets were collected through a platform developed for the European Joint Action on Vaccination. Two datasets were collected, the first between November 2019 and June 2020, the second from April to September 2021. The tweets were manually categorized by three independent annotators. After cleaning, the total dataset consisted of 1,736 tweets with 3 categories (promotional, neutral, and discouraging). The manually classified tweets were used to train and test various machine learning models. The model that classified the data most similarly to humans was XLM-Roberta-large, a multilingual version of the Transformer-based model RoBERTa. The model hyper-parameters were tuned and then the model ran five times. The fine-tuned model with the best F-score over the validation dataset was selected. Running the selected fine-tuned model on just the first test dataset resulted in an accuracy of 72.8% (F-score 0.713). Using this model on the second test dataset resulted in a 10% drop in accuracy to 62.1% (F-score 0.617), indicating that the model recognized a difference in language between the datasets. On the combined test datasets the accuracy was 70.1% (F-score 0.689). Retraining the model using data from the first and second datasets increased the accuracy over the second test dataset to 71.3% (F-score 0.713), a 9% improvement from when using just the first dataset for training. The accuracy over the first test dataset remained the same at 72.8% (F-score 0.721). The accuracy over the combined test datasets was then 72.4% (F-score 0.720), a 2% improvement. Through fine-tuning a machine-learning model on task-specific data, the accuracy achieved in categorizing tweets was close to that expected by a single human annotator. Regular training of machine-learning models with recent data is advisable to maximize accuracy.

How the Italian Twitter Conversation on Vaccines Changed During the First Phase of the Pandemic: A Mixed-Method Analysis.

In the context of the European Joint Action on Vaccination, we analyzed, through quantitative and qualitative methods, a random sample of vaccine-related tweets published in Italy between November 2019 and June 2020, with the aim of understanding how the Twitter conversation on vaccines changed during the first phase of the pandemic, compared to the pre-pandemic months. Tweets were analyzed by a multidisciplinary team in terms of kind of vaccine, vaccine stance, tone of voice, population target, mentioned source of information. Multiple correspondence analysis was used to identify variables associated with vaccine stance. We analyzed 2,473 tweets. 58.2% mentioned the COVID-19 vaccine. Most had a discouraging stance (38.1%), followed by promotional (32.5%), neutral (22%) and ambiguous (2.5%). The discouraging stance was the most represented before the pandemic (69.6%). In February and March 2020, discouraging tweets decreased intensely and promotional and neutral tweets dominated the conversation. Between April and June 2020, promotional tweets remained more represented (36.5%), followed by discouraging (30%) and neutral (24.3%). The tweets' tone of voice was mainly polemical/complaining, both for promotional and for discouraging tweets. The multiple correspondence analysis identified a definite profile for discouraging and neutral tweets, compared to promotional and ambiguous tweets. In conclusion, the emergence of SARS-CoV-2 caused a deep change in the vaccination discourse on Twitter in Italy, with an increase of promotional and ambiguous tweets. Systematic monitoring of Twitter and other social media, ideally combined with traditional surveys, would enable us to better understand Italian vaccine hesitancy and plan tailored, data-based communication strategies.

Intention of Parents to Immunize Children against SARS-CoV-2 in Italy.

Several countries have targeted adolescents for immunization against SARS-CoV-2 to mitigate COVID-19 spread. In Italy, immunization for children ≥ 12 years has been available starting from June 2021. We conducted a cross-sectional study to investigate the knowledge, attitude and intention to vaccinate children < 18 years in Italian families. We used a multinomial logistic regression model to investigate factors associated with intention to vaccinate. We collected a total of 1696 responses. Among the 491 families of children ≥ 12 years, 41.2% would not vaccinate their children and 21.2% were uncertain, while among the 1205 families of children < 12 years, 36.1% would not vaccinate and 33.8% were uncertain. Determinants of intention to vaccinate both age groups were perceived safety and efficacy of vaccines and perceived risk of transmitting infection to adults. For children < 12 years, additional determinants were perceived risk of being infected and being hospitalized because of COVID-19. In view of the expanding strategy to vaccinate adolescents and the availability of immunization for children < 12 years, our results call for a communication strategy targeted at families of children focused on the safety and efficacy of COVID-19 vaccine in children and on the dynamics of infection spread across different age groups. As perceptions in families are volatile and may change rapidly over time, repeated surveys for measuring attitudes to vaccinate would be advisable.

Digital Surveillance Through an Online Decision Support Tool for COVID-19 Over One Year of the Pandemic in Italy: Observational Study.

BACKGROUND: Italy has experienced severe consequences (ie, hospitalizations and deaths) during the COVID-19 pandemic. Online decision support systems (DSS) and self-triage applications have been used in several settings to supplement health authority recommendations to prevent and manage COVID-19. A digital Italian health tech startup, Paginemediche, developed a noncommercial, online DSS with a chat user interface to assist individuals in Italy manage their potential exposure to COVID-19 and interpret their symptoms since early in the pandemic. OBJECTIVE: This study aimed to compare the trend in online DSS sessions with that of COVID-19 cases reported by the national health surveillance system in Italy, from February 2020 to March 2021. METHODS: We compared the number of sessions by users with a COVID-19-positive contact and users with COVID-19-compatible symptoms with the number of cases reported by the national surveillance system. To calculate the distance between the time series, we used the dynamic time warping algorithm. We applied Symbolic Aggregate approXimation (SAX) encoding to the time series in 1-week periods. We calculated the Hamming distance between the SAX strings. We shifted time series of online DSS sessions 1 week ahead. We measured the improvement in Hamming distance to verify the hypothesis that online DSS sessions anticipate the trends in cases reported to the official surveillance system. RESULTS: We analyzed 75,557 sessions in the online DSS; 65,207 were sessions by symptomatic users, while 19,062 were by contacts of individuals with COVID-19. The highest number of online DSS sessions was recorded early in the pandemic. Second and third peaks were observed in October 2020 and March 2021, respectively, preceding the surge in notified COVID-19 cases by approximately 1 week. The distance between sessions by users with COVID-19 contacts and reported cases calculated by dynamic time warping was 61.23; the distance between sessions by symptomatic users was 93.72. The time series of users with a COVID-19 contact was more consistent with the trend in confirmed cases. With the 1-week shift, the Hamming distance between the time series of sessions by users with a COVID-19 contact and reported cases improved from 0.49 to 0.46. We repeated the analysis, restricting the time window to between July 2020 and December 2020. The corresponding Hamming distance was 0.16 before and improved to 0.08 after the time shift. CONCLUSIONS: Temporal trends in the number of online COVID-19 DSS sessions may precede the trend in reported COVID-19 cases through traditional surveillance. The trends in sessions by users with a contact with COVID-19 may better predict reported cases of COVID-19 than sessions by symptomatic users. Data from online DSS may represent a useful supplement to traditional surveillance and support the identification of early warning signals in the COVID-19 pandemic.

Sars-Cov2 Not Detected in a Pediatric Population With Acute Respiratory Infection in Primary Care in Central and Southern Italy From November 2019 to Early March 2020.

Background: In December 2019, a novel coronavirus named SARS-CoV-2 started circulating in China and this led to a major epidemic in Northern Italy between February and May 2020. Young children (aged <5 years) seem to be less affected by this coronavirus disease (COVID-19) compared to adults, although there is very little information on the circulation of this new virus among children in Italy. We retrospectively tested nasopharyngeal swabs for SARS-CoV-2 in samples collected in young children between November, 2019 and March, 2020 in the context of the RSV ComNet study. Methods: Two networks of primary care pediatricians in Lazio (Central Italy) and Puglia (Southern Italy) collected nasopharyngeal swabs from children, aged <5 years, presenting with symptoms for an acute respiratory infection (ARI). The RSV ComNet study is a multicenter study implemented to estimate the burden of RSV in young children (aged <5 years) in the community. Swabs were sent to a central reference laboratory and tested for 14 respiratory viruses through RT-PCR. All collected samples were retrospectively tested for SARS-CoV-2 using RT-PCR (Istituto Superiore di Sanità protocol). Results: A total of 293 children with ARI were identified in the two participating networks. The highest number of cases were recruited in weeks 51/2019 and 3/2020. The majority of patients (57%) came from the Lazio region. All of the 293 samples tested negative for SARS-Cov2. Rhinovirus was the most frequently detected virus (44%), followed by RSV (41%) and influenza viruses (14%). Conclusions: Our study shows that in Lazio (a region of intermediate SARS-COV-2 incidence) and Puglia (a region of low incidence), the SARS-Cov2 virus did not circulate in a sample of ARI pediatric cases consulting primary care pediatricians between November 2019 and March 2020.