Strategies of the Spanish Ministry of Health on Instagram: A Comparative Study Before, During, and After the COVID-19 Pandemic

Estrategias del Ministerio de Sanidad de España en Instagram: Un estudio comparativo antes, durante y después de la pandemia COVID-19

Estratégias do Ministério da Saúde Espanhol no Instagram: Um Estudo Comparativo Antes, Durante e Depois da Pandemia da COVID-19

Douglas Farias Cordeiro^1*
Carlos Lopezosa^2**
Mari Vállez^2***
Javier Guallar^2****

¹ Federal University of Goiás, Goiânia, Brasil
² University of Barcelona, Spain

^* Professor at the Faculty of Information and Communication, Federal University of Goiás, Goiânia, Brazil. Email: cordeiro@ufg.br
^** Lecturer at the University of Barcelona, Spain. Email: lopezosa@ub.edu
^***Professor in the Department of Library Science, Documentation, and Audiovisual Communication at the Faculty of Information and Audiovisual Media, University of Barcelona, Spain. Email: marivallez@ub.edu
^****Professor and Academic Secretary of the Faculty of Information and Audiovisual Media, University of Barcelona, Spain. Email: jguallar@ub.edu

Abstract

This study analyses the use of Instagram by the Spanish Ministry of Health to understand how this platform has been used to disseminate content and engage the public on health issues. Posts from 2018 and 2023 are examined, segmented into three periods: before the COVID-19 pandemic, during the pandemic, and after the pandemic. The study uses a quantitative analysis of the data to understand the volume and type of posts, engagement metrics, and the use of hashtags and mentions. An artificial intelligence model identifies the names of people mentioned in the posts. The results indicate that during the pandemic, the increase in posts and the use of videos boosted interaction, while mentions and names of individuals reduced engagement. Additionally, in all periods, shorter texts and the use of carousels and specific hashtags were associated with higher engagement, highlighting the importance of these tactics in institutional communication on social networks.

Keywords
Health; Social media; Instagram; Ministry of Health; COVID-19; Pandemics.

Resumen

Este estudio analiza el uso de Instagram por el Ministerio de Sanidad de España con el objetivo de comprender cómo se ha utilizado esta plataforma para difundir contenido y comprometer al público en temas sanitarios. Se consideran las publicaciones realizadas entre 2018 y 2023, segmentadas en tres períodos: antes de la pandemia de COVID-19, durante la pandemia y después de la pandemia. Se realiza un análisis cuantitativo sobre los datos para comprender el volumen, tipo de publicaciones, métricas de participación, y el uso de hashtags y menciones. Se utiliza un modelo de inteligencia artificial para la identificación de nombres de personas mencionados en los textos de las publicaciones. Los resultados obtenidos indicaron que durante la pandemia, el aumento de publicaciones y el uso de vídeos incrementaron la interacción, mientras que las menciones y nombres de personas redujeron la participación. Además, en todos los períodos, los textos más cortos y el uso de carruseles y hashtags específicos se asociaron con un mayor engagement, destacando la importancia de estas tácticas en la comunicación institucional en redes sociales.

Palabras clave
Salud; Redes sociales; Instagram; Ministerio de Sanidad; COVID-19; Pandemias.

Resumo

Este estudo analisa o uso do Instagram pelo Ministério da Saúde da Espanha com o objetivo de compreender como esta plataforma tem sido utilizada para disseminar conteúdo e engajar o público em temas de saúde. São consideradas as publicações realizadas entre 2018 e 2023, segmentadas em três períodos: antes da pandemia de COVID-19, durante a pandemia e após a pandemia. É realizada uma análise quantitativa dos dados para compreender o volume, tipo de publicações, métricas de participação e o uso de hashtags e menções. Utiliza-se um modelo de inteligência artificial para a identificação de nomes de pessoas nos textos das publicações. Os resultados obtidos indicaram que, durante a pandemia, o aumento das publicações e o uso de vídeos aumentaram a interação, enquanto as menções e nomes de pessoas reduziram a participação. Além disso, em todos os períodos, textos mais curtos e o uso de carrosséis e hashtags específicas foram associados a um maior engajamento, destacando a importância dessas táticas na comunicação institucional em redes sociais.

Palavras chave
Saúde; Redes sociais; Instagram; Ministério da Saúde; COVID-19; Pandemias.

1. Introduction

In the context of public institutions, the need to establish efficient and effective communication channels and strategies is paramount to ensuring the adequate dissemination of information, promoting institutional transparency, and strengthening the relationship between the institution and the citizen. In a connected and digitized society, communication is a crucial tool for building trust, commitment, and social collaboration (Lovari & Valentini, 2020).

Over the years, the Ministry of Health (MISAN) has faced significant challenges in adapting its communication to the ever-evolving needs and demands of society (Martínez-Estrella, 2020). This has involved implementing strategies to raise awareness about health issues, promote adherence to prevention programs, and provide clear and accessible information about available healthcare services (Pulido-Polo et al., 2021; Forja-Pena, 2022). The adoption of digital platforms such as Instagram, Facebook, X (formerly Twitter), and TikTok has become essential to reach a broader and more diverse audience, particularly during times of crisis, such as the COVID-19 pandemic (Abuín-Penas & Abuín-Penas, 2022; Forja-Pena, 2022; Ignacio-Criado et al., 2020).

The use of Instagram as a communication channel by MISAN began in 2018 through the official profile @sanidadgob. Instagram is a social media platform specializing in image and video posts, which can include descriptive texts and location information (Pittman & Reich, 2016). The platform offers various forms of interaction, such as comments, shares, and “likes,” encouraging user engagement with content. Additionally, the use of hashtags is a common practice for indexing and organizing content, facilitating the discovery of related posts. Similarly, the ability to mention other users in posts creates a dynamic of connection and direct communication on the platform, enhancing debate, information dissemination, and the establishment of implicit communities. In recent years, Instagram has been extensively explored as a platform for public institution communication due to its popularity, especially among younger populations (Leone & Della Mura, 2020).

In an environment of constant Instagram usage growth, exploring communication strategies focused on health promotion and the provision of reliable information is of great importance and requires continuous action and planning by responsible entities, not only in crisis situations (Pinto et al., 2020). This article proposes an analysis of information circulation on MISAN’s official Instagram profile during three periods: before the COVID-19 pandemic (August 18, 2018, to March 10, 2020), during the pandemic (March 11, 2020, to May 4, 2023), and after the pandemic (May 5 to December 31, 2023). The study is a quantitative investigation of how MISAN used the platform to disseminate content to society. By examining the evolution of posts, engagement, and message dynamics across these three periods, the study aims to understand not only Instagram’s role as an informational channel but also the strategies adopted to inform and engage the public on health issues.

The research focuses on using statistical, computational, and data visualization strategies, emphasizing the following indicators: analysis of publication volume and engagement, hashtag and mention network analysis, and thematic identification and analysis. The defined research objectives are as follows:

O1: Analyze MISAN’s use of Instagram in terms of volume and types of posts and content.
O2: Analyze engagement metrics (likes and comments).
O3: Compare the data across the three analyzed periods (pre-pandemic, pandemic, and post-pandemic), considering engagement segmentation by quartiles.

2. Theoretical framework

The COVID-19 pandemic highlighted the importance of digital communication, particularly on social media, as a platform for exchanging thoughts and positions related to the crisis, transcending geographical boundaries (Xifra, 2020). This has led to the emergence of community initiatives and a reevaluation of professional roles in community health action, emphasizing the need for effective communication strategies (Castillo-Esparcia et al., 2020).

Some of these strategies have been conceptualized as e-health (Díaz-García & Girón-Prieto, 2022), referring to actions involving the application of information and communication technologies in the field of health (Tebej & Klein, 2021). In this context, social networks and virtual communities have taken on a fundamental role in e-health (Díaz-García & Girón-Prieto, 2022), sparking growing research interest in recent years, particularly focused on platforms such as X (formerly Twitter) and the COVID-19 pandemic.

A relevant precedent is the study by Cano Garcinuño and Arce-García (2020), which analyzed social media campaigns aimed at promoting influenza vaccination during 2018. Subsequently, research has focussed on the COVID-19 pandemic.

Previous studies on the social network X have addressed several key aspects. On the one hand, the communication strategies of public health authorities in Mexico and Spain during the COVID-19 health crisis have been analyzed by reviewing official profiles on X to determine the narratives created and the use of metaphors in their communication (Martínez-Estrella, 2020). On the other hand, the types of content shared about COVID-19 on the official X account of the Spanish Ministry of Health from March to December 2020 have been extensively studied, confirming significant differences in content publication during different stages of the pandemic (Tenorio and Gómez-Carmona, 2021). Additionally, the effectiveness of crisis communication during the second wave of COVID-19 has been examined through critical discourse analysis of more than 500 tweets from the X accounts of the Spanish government and the Ministry of Health (Poch-Butler & Martínez, 2021).

Thus, studies on e-health using the X platform demonstrate the potential of this social network as an effective online institutional communication tool during crises, based on informational transparency and a constant flow of information (Pulido-Polo et al., 2021). Consequently, it has been observed that communication on X has been essential for governance and citizen preparedness during crises, showing that well-executed strategies by official organizations are effective in addressing pandemics (Nicasio-Varea et al., 2023).

Similarly, other studies have addressed additional social networks, with a particular focus on ministerial communication on TikTok. Research has analyzed the level of engagement of health ministries in Spain, the United Kingdom, and Germany on this platform to identify communication strategies and the types of content generating the most interaction (Forja-Pena, 2022). Another study examined how seven Spanish government ministries use TikTok to communicate with citizens, evaluating content types, formats, tone, and posting frequency. This study revealed a lack of coherence in the Spanish government’s communication strategy for optimizing its presence on this platform (Alonso-López et al., 2024).

The “Este virus lo paramos unidos” (“Together, we stop this virus”) campaign by the Ministry of Health was analyzed, focusing on material aimed at the general public on its website during March, April, and May 2020. This revealed that most messages were accurate, rigorous, clear, and easy to understand, thereby contributing to effective communication about COVID-19 (Huertas-Ciórraga, 2021).

3. Materials and methods

This study constitutes a quantitative, descriptive, and comparative investigation into the use of Instagram by the Spanish Ministry of Health (@sanidadgob). The methodology is based on data analysis and employs computational solutions such as web scraping, descriptive statistics, artificial intelligence, and data visualization. The scope of the research covers all posts made on the @sanidadgob profile between August 18, 2018 (date of the profile’s first post) and December 31, 2023.

To understand the strategies employed during the three distinct periods—pre-pandemic (August 18, 2018, to March 10, 2020), pandemic (March 11, 2020, to May 4, 2023), and post-pandemic (May 5 to December 31, 2023)—the following attributes were examined: number of posts, engagement, topics, profile mentions, hashtags, and personal names. The periods are based on declarations by the World Health Organization (WHO). The dataset comprises a total of 1,412 posts. The results are divided into these three periods, initially addressed from a macro perspective focused primarily on describing statistical indicators and subsequently adopting a comparative view aimed at establishing correlations and implications of the mechanisms and strategies used by MISAN.

Data was collected using a computational web scraping solution (Mitchell, 2024). A data extraction robot was developed using Python libraries Selenium and BeautifulSoup. First, the robot mapped all posts from the profile, extracting unique identifiers (IDs)—a combination of letters and numbers that allows access to a specific post. In Instagram, a post’s URL consists of the platform’s address followed by “/p/[identifier code].” The list of all post IDs from the profile of interest was stored in a temporary file, which served as input for the second phase of data extraction procedures. During the second phase, the robot accessed each post individually and analyzed the HTML (Hypertext Markup Language) code to extract relevant data, which was consolidated into a semi-structured CSV (comma-separated values) file.

The exploratory analysis aimed to identify general indicators and visualize temporal fluctuations in posts using statistical measures, data summaries, and information visualization. The following indicators were considered: total volume, daily average volume (ratio of total volume to the number of days analyzed per period), daily averages of “likes” and comments per post, and daily recurrence (ratio of days with at least one post to the total number of days). To understand patterns associated with interaction trends, indicators related to quartiles and the volume of outliers were also calculated. Outliers were identified using the interquartile range heuristic (Dash et al., 2023).

Regular expression-based computational routines (Nield, 2019) were used to identify occurrences of hashtags and mentions. Regular expressions are formal linguistic patterns used to automatically identify, extract, and manipulate textual data. For hashtags, the pattern “#\w+” was defined, where “#” matches the hashtag character, and “\w+” corresponds to any sequence of text-based characters, including letters, numbers, and the underscore character (“_”). Similarly, mentions were identified using the pattern “@\w+.” The identified hashtags and mentions were consolidated into secondary files linked to the main dataset via the post identifier attribute (ID).

Additionally, named entity recognition (NER), based on a pre-trained BERT model (Bidirectional Encoder Representations from Transformers) (Devlin et al., 2018), was applied to identify personal names in the textual data from posts. A computational routine was built to translate the original text into English using the Googletrans API and process the text with the pre-trained BERT-based NER model. Translation was necessary because the pre-trained model was optimized for English.

NER is a natural language processing (NLP) technique that automatically identifies and classifies entities in unstructured text, including people, organizations, and places (Li et al., 2022). Various algorithmic strategies can be used for NER, with BERT-based solutions being particularly effective due to their architecture, which allows bidirectional text processing, considering the context of a word based on both preceding and succeeding words. This improves the accuracy of entity identification in linguistically complex or contextually ambiguous texts (Casola et al., 2022).

Distinguishing between user mentions and personal name mentions, while potentially referring to the same individual in a single post, can impact engagement metrics. Names typically appear in explanatory or contextual text, whereas user mentions are interactive tools that encourage direct participation, allowing users to access the mentioned profile and potentially increasing visibility and engagement. Analyzing these variables separately helps identify their contributions to interaction dynamics on the platform.

Finally, the results provide a general overview of the posts, considering statistical indicators related to descriptive and engagement attributes. For each period, data were analyzed based on quartile segmentation of engagement, considering posts with engagement below the first quartile, posts between the first and third quartiles (interquartile range), and posts with engagement above the third quartile (upper quartile). Visualizations were generated for the most frequent terms (similarity graph) and a summarized view of descriptive statistical indicators.

4. Results

4.1 General Data Analysis

Table 1 presents general data considering the three time periods. Significant differences in total volume are expected, given that the temporal intervals for each period differ, requiring an analysis focused on data proportionality.

For the pre-pandemic period, the total volume is 313 posts, with a daily frequency of 38.61%. This can be interpreted as approximately one post every three days, assuming a continuous projection. In contrast, the pandemic and post-pandemic periods featured 893 and 206 posts, respectively, with daily frequencies of 47.74% and 51.05%. This corresponds to an approximate projection of one post every two days for both periods.

These indicators point to an increase in the frequency of content delivery by MISAN on its Instagram profile.

Referring to Table 1, unlike the daily recurrence indicator, which measures the percentage of days on which content was published, the daily average volume captures nuances derived from quantitative variations in the total number of posts.

It is noteworthy that the pre-pandemic and post-pandemic periods showed average values below 1. Combined with the daily recurrence of these periods, this indicates that single-post days predominated. On the other hand, the pandemic period presented a daily average volume of 1.55, indicating more than one post per day. It is important to highlight that for both the pandemic and pre-pandemic periods, the coefficient of variation exceeded 100%, signaling the presence of days with two or more posts.

In terms of audience engagement, the pre-pandemic period exhibited an average of 136.99 likes and 9.96 comments per post, with high coefficients of variation, nearing 700% for both indicators. This indicates significant dispersion in engagement, with posts showing low interaction alongside others with high engagement volumes. The values observed during the pandemic period are more than ten times higher than those of the previous period. Additionally, a notably lower coefficient of variation indicates more consistent engagement levels. In the post-pandemic period, engagement indicators are approximately half of those observed during the pandemic, with coefficients of variation nearly double those from the pandemic period. This suggests a higher engagement level than in the pre-pandemic period but with greater variability.

Table 2 presents statistical indicators related to audience engagement across the three analyzed periods. The daily proportional engagement, calculated from the cumulative average of likes and comments, shows significant variation between periods, with values of 146.96 during the pre-pandemic period, 1,451.81 during the pandemic, and 646.24 in the post-pandemic period.

Additionally, the coefficient of variation for daily proportional engagement, which indicates the relative variability of the data compared to its mean, is provided. The coefficients of variation are 636.34% for the pre-pandemic period, 132.57% for the pandemic period, and 213.57% for the post-pandemic period. Notably, the higher the coefficient of variation, the greater the data variability relative to the mean. These values can be interpreted as indicative of the magnitude of variation in engagement for each period relative to the observed average.

The coefficient of variation of 636.34% in the pre-pandemic period indicates a relatively high dispersion of data in relation to the mean during that time. In contrast, the coefficient of variation of 132.57% during the pandemic period suggests lower data dispersion relative to the mean, while the 213.57% coefficient of variation in the post-pandemic period reflects an intermediate level of dispersion. These results suggest a substantial increase in user engagement with MISAN’s Instagram posts during the pandemic period, followed by a percentage decrease in the post-pandemic period, though engagement levels remained higher than those observed in the pre-pandemic period.

The quartile values presented in Table 2 provide insights into the distribution of engagement data. The first quartile, representing the threshold below which 25% of the records fall, showed a significant increase during the pandemic period compared to the pre-pandemic period. While the first quartile value was 35.00 in the pre-pandemic period, this jumped to 383 during the pandemic, indicating that 75% of the posts achieved engagement values above this threshold. A similar trend is observed in the upper quartile: during the pre-pandemic period, the value was 81.00, whereas during the pandemic it reached 1,838.00, representing a growth of 2,167%. Comparing the pandemic period to the post-pandemic period, which had an upper quartile value of 466.25, there is a percentage decrease of 74.62%. Despite this decline in the third period, the values remain higher than those observed in the pre-pandemic period.

Another notable aspect of engagement patterns is the discrepancy between minimum and maximum values, as shown in Table 2. These values reach orders of thousands across all analyzed periods, underscoring the extent of variability in user interaction with MISAN’s Instagram posts. Furthermore, when considering indicators related to outliers, there is significant variance between the coefficient delimiting the threshold for upper outliers and the maximum values. This indicates that even among outlier values, certain posts achieved particularly high engagement, highlighting their exceptional impact.

4.1.1 Type of Content

Table 3 presents indicators related to media type and the use of the carousel format. During the pre-pandemic period, the use of images significantly exceeded that of videos. However, during the pandemic, this trend reversed, with videos representing the highest percentage of content—a trend that persisted in the post-pandemic period. The use of social media to share videos featuring excerpts from official communications on health topics has become a well-established and publicly engaging practice (Castillo-Esparcia et al, 2020).

The percentage of use of the carousel format was close to 20% during the pre-pandemic and post-pandemic periods, while it dropped to 9.18% during the pandemic. The carousel format allows for the dissemination of a larger amount of information in a more interactive manner, especially compared to the use of long text captions in posts.

4.1.2 Collaborative Posts

Since October 19, 2021, Instagram has allowed content to be associated with multiple profiles, a feature known on the platform as “collab mode” (Instagram, 2021). During the pandemic period, only one collaborative post was identified, while six such posts were identified in the post-pandemic period, all published in 2023.

The first collaborative post, still within the pandemic period, was published with the @desdelamoncloa profile regarding the end of mandatory mask use on public transportation. This post achieved an engagement of 1,066.00, while posts without collaborative mode during the same period showed a higher average engagement of 1,452.24.

In the post-pandemic period, a collaborative post was observed from the Ministry of Culture profile (@culturagob) regarding #OrgulloLGTBI, achieving an engagement of 281.00. Additionally, five collaborative posts were published in December 2023 from the official profile of the Minister of Health, Mónica García Gómez (@monicagarciag_), highlighting actions undertaken by MISAN. These posts achieved an average engagement of 6,700.40, compared to an average engagement of 496.71 for non-collaborative posts during the same period.

It is important to note that although the number of collaborative posts from MISAN’s profile is low, this type of content contributes organically to expanding and diversifying its reach, as the posts engage the followers of other profiles.

4.1.3 Hashtag Analysis

Regarding the use of hashtags, during the pre-pandemic period, 689 occurrences were identified across a total of 248 posts containing hashtags, indicating the use of this marker in 79.23% of the posts. A more detailed analysis, presented in Table 4, shows that the most frequently used hashtag was #Repost, which signifies the reposting of content previously shared on other profiles.

Interestingly, the percentage values for each of the top five hashtags are relatively low and close to one another, suggesting that there was no concentration in the use of specific hashtags. Additionally, the use of generic hashtags, such as #sanidad (healthcare) and #salud (health), was observed.

It is important to note that hashtags serve as indexing mechanisms on social media, facilitating the search for content of interest. However, the use of broad and generic hashtags may have limited impact on the organic reach of content delivery (Bruns & Burgess, 2011).

During the pandemic period (Table 4), the hashtag #COVID19 was mentioned 277 times, representing 12.43% of the total hashtags identified during this period. This indicates a significant focus on pandemic-related issues and public health measures. Notably, the top five hashtags revolve around pandemic-related topics, with a particular emphasis on the hashtag #EsteVirusLoParamosUnidos, referencing the campaign of the same name that promoted the dissemination of information and guidance on COVID-19 (Villodre & Criado, 2021). This strategy goes beyond a generalist approach, adding meaning and context beyond the indexing function.

During the pandemic period, 2,229 hashtags were identified across 774 posts, representing a 43.33% usage rate relative to the total posts in the period.

In the post-pandemic period, the proportional use of hashtags was similar to the pre-pandemic period, with 342 hashtags identified in 164 posts containing hashtags, representing 79.61% usage. The increased thematic diversity of the main hashtags (Table 4) was noted in this period, which included topics such as mental health, HIV, and public policy. However, similarly to the other periods, there was a low frequency of hashtag use overall, reflecting a less consistent and associative pattern in leveraging hashtags as a strategy for organic content reach.

4.1.4 Analysis of Personal Names

The identification of personal names, understood as explicit references to individuals’ names rather than user profiles (which will be analyzed in section 4.1.5), revealed notable use of this strategy by MISAN, particularly concerning the successive ministers leading the department.

During the pre-pandemic period, 45 different names were identified across a total of 136 name occurrences within 80 posts, representing 25.55% of posts. As shown in Table 5, the most frequently mentioned name was that of physician and politician María Luisa Carcedo, who served as Minister of Health from September 2018 to January 2020. Her name accounted for 38.97% of occurrences. Carcedo was succeeded by Salvador Illa in January 2020.

In the pandemic period, a considerably lower percentage of posts (16.18%) included mentions of personal names. In this period, 282 names were identified across 520 mentions. The most frequently mentioned name was that of Carolina Darias, who served as Minister of Health from January 2021 to March 2023. Although Salvador Illa also served as Minister during the pandemic (from January 2020 to January 2021), his name appeared in only four posts, representing 0.77% of mentions—similar to mentions of the Spanish diplomat Guillermo Kirkpatrick, as noted in Table 5.

Finally, in the post-pandemic period, 45.63% of posts included mentions of personal names, with a total of 85 names identified in 203 mentions. Following the pattern observed in previous periods, the name of then-Minister José Manuel Miñones was the most frequently mentioned, accounting for nearly half of all mentions. This indicates a strong association between Miñones and MISAN in communications through its Instagram profile.

4.1.5 Analysis of Profile Mentions

Mentions of other profiles on Instagram are a resource that establishes a direct connection with the mentioned profile, allowing other users to be included in a post and enhancing interactions.

In the pre-pandemic period, 203 mentions were observed across a total of 145 posts, representing the use of this strategy in 46.32% of all posts during this period. As shown in Table 6, the most frequently mentioned profile was that of Minister Luisa Carcedo (@carcedoluisa).

During the pandemic period, the proportional use of mentions decreased significantly, with only 12.15% of posts including mentions. This accounted for a total of 217 mentions across 1,786 posts. Similar to the pre-pandemic period, the most mentioned profile was that of Minister Carolina Darias (@carolinadarias), representing 30.00% of all mentions. The second most mentioned profile, though at a considerable distance, was that of Minister Salvador Illa (@salvador_illa), accounting for 4.62% of mentions.

For the post-pandemic period, a proportional increase was observed compared to the pandemic period, although the number of mentions remained lower than in the pre-pandemic period. A total of 95 mentions were identified across 64 posts out of 206, representing 31.06% of posts during this period.

One notable aspect is the percentage of mentions of Minister José Miñones (@minones), which accounted for approximately half of all mentions. Additionally, though with a small percentage of overall mentions, it is worth highlighting the self-mention of the Ministry of Health’s profile in all the analyzed periods, consistently ranking among the top five most mentioned profiles (Table 6).

4.2 Segmented Engagement Analysis

Shifting focus to the relationship between the strategies adopted by MISAN and the potential of Instagram as a communication channel, Table 7 presents a view of correlation coefficients with respect to engagement values. The indicators in Table 7 individually represent the percentage influence of each attribute.

For instance, the attribute “Mention” during the pre-pandemic period showed a value of -6.98%, indicating that as the use of mentions in posts increased, a decrease in user engagement was observed.

Table 7 reveals that, in terms of direct mentions via user profile tags as well as mentions by personal name, the percentage influence on engagement is negative and has shown a temporal increase across the analyzed periods.

Regarding hashtags, a positive influence was identified only during the pandemic period, with a percentage value of 21.17%. Conversely, the pandemic period also saw the most significant negative influence associated with the use of videos. It is worth noting that for other periods, the influence of videos on engagement was almost negligible, with values close to zero.

The length of captions associated with posts also showed negative influences, indicating that shorter captions were associated with higher engagement levels. Finally, the use of the carousel format emerged as the most strategically impactful indicator, showing a positive percentage increase in engagement, rising from 3.31% during the pandemic to 8.04% in the post-pandemic period.

The following subsections present segmented indicators by period and engagement levels in terms of quartiles, to better understand the influences associated with different strategies used by MISAN on Instagram. This approach allows the identification of possible patterns associated with lower engagement (first quartile), median engagement (interquartile range), or higher engagement (upper quartile).

4.2.1 Pre-pandemic

Table 8 presents the comparison of indicators for the pre-pandemic period. The percentage indicators related to the use of mentions, occurrences of personal names, and video usage showed no substantial differences across the three segments.

Regarding the use of hashtags, posts with median engagement (interquartile range) exhibited the highest percentage value. However, when analyzing posts with the highest engagement (upper quartile), the indicator revealed a lower relative percentage of hashtag usage.

The most significant indicator was the length of the text in the posts’ captions, with an average of 300.32 characters for the first quartile and 261.56 characters for the upper quartile. This suggests that posts with shorter captions tended to achieve higher engagement levels.

4.2.2 During the Pandemic

During the pandemic period, slightly different patterns emerge compared to the pre-pandemic period. Regarding the use of mentions and the occurrence of personal names, Table 9 indicates that increased engagement is strongly associated with a decrease in the use of these strategies. For example, in the first quartile, the percentage of posts featuring occurrences of personal names is 47.98%, whereas for the upper quartile, this percentage drops to 13.90%.

In terms of hashtags, a slight increase in usage is observed in relation to engagement. The use of videos and carousels showed no significant variations during the pandemic period.

Finally, as was observed in the pre-pandemic period, shorter captions are associated with posts that achieve higher engagement levels.

4.2.3 Post-Pandemic

For the records corresponding to the post-pandemic period, Table 10 indicates that higher engagement (upper quartile) is associated with a lower use of mentions and fewer occurrences of personal names. However, in this case, the percentage difference is smaller than what was observed for posts during the pandemic period.

Additionally, the upper quartile also showed a higher percentage of posts using hashtags and the carousel format compared to posts in the first quartile and interquartile range.

The use of videos follows an inversely proportional trend to engagement, meaning that image-based posts are more prevalent among those with higher engagement.

In terms of caption length, a similar pattern to previous periods is observed, where posts with longer captions tend to fall within the lower engagement range (first qartile).

5. Discussion and Conclusions

The analysis of the communication strategies of the Spanish Ministry of Health on Instagram shows a significant evolution in the frequency and type of posts, as well as in engagement metrics, across the three analyzed periods: pre-pandemic, pandemic, and post-pandemic. During the pandemic, the increase in posting frequency and the use of videos led to higher interaction levels, accompanied by a notable rise in engagement variability. In this phase, one of Instagram’s main features—its focus on sharing visual content such as images and videos—was strongly leveraged (Ardèvol et al., 2021).

Regarding the first objective, the use of hashtags had a positive influence, particularly during the pandemic, reflecting a focus on disseminating specific information and campaigns related to COVID-19. Mentions of the health ministers from each of the three periods, as well as their Instagram profiles, were the most frequently referenced personal entities in the content. An exception to this pattern was observed in references to Salvador Illa’s name, which might be related to a strategy more focused on disseminating non-personalized information during Illa’s tenure as Minister of Health. This issue warrants further exploration in future research.

For the second objective, the analysis of engagement metrics for the posts revealed different patterns across the pre-pandemic, pandemic, and post-pandemic periods. During the pre-pandemic period, the average number of “likes” and comments per post was relatively low, at 136.99 and 9.96 respectively, with high variability in interaction. However, during the pandemic, these metrics increased significantly, with a much higher average of likes and comments (1,451.81 and 35.32 respectively), indicating greater user interest and content relevance. In the post-pandemic period, although the levels of “likes” and comments decreased compared to the pandemic, they remained above pre-pandemic levels, establishing a more engaged follower base. Furthermore, posts with higher engagement tended to use content strategies that included specific hashtags, such as COVID-19, and mentions of relevant profiles, highlighting the importance of these tactics in maximizing engagement.

Lastly, for the third research objective, the posts by the Spanish Ministry of Health (MISAN) during the three analyzed periods revealed some differences between these phases. In the pre-pandemic period, the use of mentions and videos had a negative influence on public engagement, while shorter captions proved to be more effective. During the pandemic, although mentions and references to individuals continued to show a similar trend in terms of engagement, the use of hashtags had a positive effect, and carousel posts also began to yield better results. In the post-pandemic period, mentions and references to individuals remained less effective, while hashtags and carousel posts maintained their positive influence on engagement. Overall, posts with shorter texts and the use of carousels were associated with higher engagement across all analyzed phases.

In conclusion, in an increasingly digitized world, the ability of public institutions to communicate effectively through social media is key to strengthening citizen trust and engagement. This research analyzes the use of Instagram by a public institution during a public health emergency, comparing it with the periods before and after. This information is highly relevant as it examines institutional communication on one of the most popular social networks, particularly among young people—Instagram—over an extended period of approximately five years (2018–2023), during which one of the most severe public health emergencies of recent decades occurred.

The limitations of this study lie in the absence of an in-depth analysis of the textual content, which could have provided insights into which topics were emphasized, such as whether they focused on public health issues or those related to the ensuing economic crisis. Additionally, conducting comparative studies with other countries could offer a more global and comprehensive perspective on official social media communication strategies during health crises. Both aspects can be explored and are therefore suggested for future research.

Author’s Contribution

Douglas Farias Cordeiro: conceptualization, investigation, methodology, data collection, analysis, visualization, writing - original draft, writing- review and editing. Carlos Lopezosa: conceptualization, investigation, writing - original draft, writing- review and editing. Mari Valléz: conceptualization, writing - original draft, writing- review and editing. Javier Guallar: conceptualization, investigation, writing - original draft, writing- review and editing. All authors have read and approved the published version of the manuscript.

Conflicts of Interest

The authors declare no conflict of interest.

Funding

This work is part of the Project “Parameters and strategies to increase the relevance of media and digital communication in society: curation, visualisation and visibility (CUVICOM)” funded by MICIU/AEI/PID2021-123579OB-I00 and by “ERDF/EU”.

References

Abuín-Penas, Javier, & Abuín-Penas, Rocío (2022). Redes sociales y el interés por la información oficial en tiempos de pandemia: análisis de la comunicación de los ministerios de salud europeos en Facebook durante la COVID-19. Revista de Comunicación y Salud, 12, 59-76. https://doi.org/10.35669/rcys.2022.12.e303

Alonso-López, Nadia, Sidorenko Bautista, Pavel, & Ferruz González, Sonia Aránzazu (2024). Administraciones públicas en TikTok. Comunicación, narrativa y frecuencia de publicación de los perfiles de los ministerios de España. Revista de Comunicación, 23(1), 33-51. https://doi.org/10.26441/RC23.1-2024-3451

Ardèvol, Elisenda, Martorell, Sandra, & San-Cornelio, Gemma (2021). Myths in visual environmental activism narratives on Instagram. Comunicar, 29(68), 59-70. https://doi.org/10.3916/C68-2021-05

Bruns, Axel, & Burgess, Jean (2011). The use of Twitter hashtags in the formation of ad hoc publics. In Axel Bruns & Pieter De Wilde (Eds.), Proceedings of the 6th European Consortium for Political Research 1 (ECPR) (pp. 1-9). ECPR, United Kingdom.

Cano Garcinuño, Maria Isabel, & Arce García, Sergio (2020). Análisis de la comunicación en redes sociales de la campaña de la vacuna de gripe en España. Revista Española de Salud Pública, 94, 202003008. Epub 02 de noviembre de 2020. Recuperado en 02 de octubre de 2024, de http://scielo.isciii.es/scielo.php?script=sci_arttext&pid=S1135-57272020000100063&lng=es&tlng=es.

Casola, Silvia, Lauriola, Ivano, & Lavelli, Alberto. (2022). Pre-trained transformers: an empirical comparison. Machine Learning with Applications, 9, 100334. https://doi.org/10.1016/j.mlwa.2022.100334

Castillo-Esparcia, Antonio, Fernández-Souto, Ana-Belén, & Puentes-Rivera, Iván (2020). Comunicación política y Covid-19. Estrategias del Gobierno de España. Profesional de la Información, 29(4), e290419. https://doi.org/10.3145/epi.2020.jul.19

Dash, Ch. Sanjeev Kumar, Behera, Ajit Kumar, Dehuri, Satchidananda, & Ghosh, Ashish (2023). An outliers detection and elimination framework in classification task of data mining. Decision Analytics Journal, 6, 100164. https://doi.org/10.1016/j.dajour.2023.100164

Devlin, Jacob, Chang, Ming-Wei, Lee, Kenton, Toutanova, Kristina (2018). BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding. arXiv:1810.04805. https://doi.org/10.48550/arXiv.1810.04805

Díaz-García, Raquel Maria, & Girón-Prieto, María Sierra (2022). Uso de redes sociales en comunicación sanitaria. FMC-Formación Médica Continuada en Atención Primaria, 29(3), 18-27. https://doi.org/10.1016/j.fmc.2022.03.003

Forja-Pena, Tania. (2022). Engagement de los ministerios de sanidad español, inglés y alemán en TikTok. Universitas-XXI, Revista de Ciencias Sociales y Humanas, (36), 173-194. https://doi.org/10.17163/uni.n36.2022.07

Huertas-Ciórraga, Elisa (2021). Análisis de los materiales de información a la ciudadanía de la campaña del Ministerio de Sanidad “Este virus lo paramos unidos” publicados entre marzo y mayo de 2020. Revista Española de Comunicación En Salud, 12(2), 121-134. https://doi.org/10.20318/recs.2021.5750

Ignacio-Criado, J., Guevara-Gómez, Ariana, & Villodre, Julián (2020). Using collaborative technologies and social media to engage citizens and governments during the COVID-19 Crisis. The Case of Spain. Digital Government: Research and Practice, 1(4),1-7. https://doi.org/10.1145/3416089

Instagram. [@creators] (2021, 19 de octubre). Today we’re launching a new feature called Collabs […] [Post]. Instagram. https://www.instagram.com/p/CVOKMVNARZX/

Leone, Stefania, & Della Mura, Miriam (2020). Institutional Flows of Communication for Young People on Instagram and the Use of Visual Images. Partecipazione & Conflitto, 13(2), 1085-1109. https://doi.org/10.1285/i20356609v13i2p1085

Li, Jing, Sun, Aixin, Han, Jianglei, & Li, Chenliang (2022). A Survey on Deep Learning for Named Entity Recognition. IEEE Transactions on Knowledge and Data Engineering, 34(1), 50-70. https://doi.org/10.1109/TKDE.2020.2981314

Lovari, Alessandro, & Valentini, Chiara (2020). Public Sector Communication and Social Media: Opportunities and Limits of Current Policies, Activities, and Practices. In Vilma Luoma-aho & María-José Canel (Eds.), The Handbook of Public Sector Communication. Opportunities and Limits of Current Policies, Activities, and Practices. John Wiley & Sons. https://doi.org/10.1002/9781119263203.ch21

Martínez-Estrella, Eva Citlali (2020). Uso de personajes y metáforas en la gestión de la crisis sanitaria del COVID-19. Revisión de la comunicación de sanidad pública en España y México. Revista Española de Comunicación en Salud, 319-327. https://doi.org/10.20318/recs.2020.5458

Mitchell, Ryan (2024). Web Scraping With Python: Data Extraction from the Modern Web. O’Reilly Media.

Nield, Thomas (2019). An Introduction to Regular Expressions. O’Reilly Media.

Pinto, Pâmela. A., Antunes, Maria João L., & Almeida, Ana Margarida P. (2020). Instagram as a communication tool in public health: A systematic review. In Proceedings of the 15th Iberian Conference on Information Systems and Technologies (CISTI) (pp. 1-6). Seville, Spain.

Pittman, Matthew, & Reich, Brandon (2016). Social media and loneliness: why an Instagram picture may be worth more than a thousand Twitter words. Computers in Human Behavior, 62, 155-167. https://doi.org/10.1016/j.chb.2016.03.084

Poch-Butler, Santana Lois, & Puebla Martínez, Belén (2021). La comunicación institucional en España en tiempos de COVID-19. Estudio de caso de las cuentas en Twitter del Gobierno de España y el Ministerio de Sanidad. In Raquel Vinader-Segura & Belén Puebla Martínez (Eds.), Ecosistema de una pandemia: COVID 19, la transformación mundial (pp. 653-676). Dykinson, Madrid.

Pulido-Polo, Marta, Hernández-Santaolalla, Víctor, & Lozano-González, Ana-Alicia (2021). Institutional use of Twitter to combat the infodemic caused by the Covid-19 health crisis. El Profesional de La Información, 30(1), 1-15. https://doi.org/10.3145/epi.2021.ene.19

Tebej, Tsion H., & Klein, Jorn (2021). Applications of e-Health to Support Person-Centered Health Care at the Time of COVID-19 Pandemic. Telemedicine and e-Health, 27(2), 150-158. https://doi.org/10.1089/tmj.2020.0201

Tenorio, Rafael Cano, & Gómez-Carmona, Diego (2021). Análisis de los tweets emitidos por parte del ministerio de sanidad de España desde la declaración oficial como pandemia de la covid-19. In Belén Puebla-Martínez & Raquel Vinader-Segura (Eds.), Ecosistema de una pandemia: COVID 19, la transformación mundial (pp. 775-796). Dykinson.

Nicasio-Varea, Blanca, Pérez Gabaldón, M., & Chavez, Manuel (2023). Estrategias de comunicación divergentes: entre el rigor y la manipulación. Análisis comparativo de la comunicación política de Estados Unidos y España frente a la etapa inicial de la crisis del COVID-19. Revista Latina de Comunicación Social, (81), 275-296. https://doi.org/10.4185/RLCS-2023-1899

Villodre, Julián, & Criado, J. Ignacio (2021). Colaboración entre administraciones públicas y ciudadanía en Twitter durante la crisis del Covid-19 en España. Gestión y Análisis de Políticas Públicas, (26), 81-91. https://doi.org/10.24965/gapp.i26.10845

Xifra, Jordi (2020). Corporate communication, public relations and reputational risk management in the days of covid-19. Profesional de la Información, 29(2), e290220. https://doi.org/10.3145/epi.2020.mar.20