Bibliometric analysis has become a valuable quantitative tool for systematically mapping the development and structure of academic fields. By examining publication trends, citation networks, keyword co-occurrences and author collaborations, bibliometrics provides replicable insights into thematic priorities and research dynamics (Aria & Cuccurullo 2017:962). Several recent studies have applied bibliometric methods to adjacent domains, including cybersecurity (Khurana et al. 2024:202; Li, Zhang & Li 2025), data breaches (Hamid & Huda 2025:1), data justice (Draude et al. 2022), digital ethics (Guenduez, Walker & Demircioglu 2025:1) and ethical artificial intelligence (Saheb & Saheb 2024:1). However, few have focused specifically on privacy issues affecting vulnerable populations, and even fewer have critically interrogated how the academic visibility of these groups reflects broader power relations in digital policy, technology development and governance. A review of recent bibliometric analyses in the privacy domain further highlights this gap. For instance, Benedetto and Cucchi (2022) traced the historical evolution of privacy concepts across five distinct eras of technological development. The study emphasises how disciplinary shifts mirror broader trends in information technology. Their work in management and information systems identified dominant themes and theoretical frameworks in privacy research but noted the lack of attention to consumer perspectives and diverse contexts. Similarly, Rauf et al. (2024) analysed privacy issues in big data applications such as finance, e-learning and healthcare. The study revealed dominant concerns around blockchain, security and technological adoption, yet overlooked how these issues affect vulnerable or marginalised users. Other studies (Ali, Zaaba & Singh 2024:863; Shu & Liu 2021:721) explored consumer privacy, digital trust and computer privacy trends across disciplines and geographies, but none addressed the socio-cultural or ethical implications of privacy for excluded populations. Van Dijk et al. (2023:81) provided one of the most comprehensive structural mappings of the privacy research field using network and topic modelling approaches, but their findings primarily emphasise dominant academic clusters (e.g. cloud computing, location privacy) while marginalising medical or ethical communities and failing to analyse which social groups remain invisible in the literature. Collectively, these studies underscore the utility of bibliometric methods in identifying key research directions, but also reveal a shared limitation: the lack of attention to epistemic inclusion and the representation of vulnerable populations in digital privacy scholarship. This shortcoming further justifies the present study’s aim to systematically analyse how different marginalised groups, particularly those affected by intersecting social, political and technological vulnerabilities, are represented (or overlooked) in privacy-related academic discourse. One notable exception in the existing body of work is the systematic literature review conducted by Sannon and Forte (2022), which explicitly focused on the intersection of privacy and marginalisation. Their study examined publications from 2010 to 2020 across human-computer interaction (HCI), communication and privacy-focused venues and proposed the privacy responses and costs framework to articulate the complex trade-offs and harms experienced by marginalised individuals. While the review made critical contributions by surfacing thematic gaps and methodological shortcomings such as the underrepresentation of race and the lack of reporting on research practices, it was limited in both temporal scope and analytical breadth (Sannon & Forte 2022). The review spanned only a decade and did not use bibliometric tools to map author networks, keyword trends, geographic distribution or institutional collaborations. Furthermore, its disciplinary scope was primarily situated within HCI and communication, omitting other fields like law, policy and information systems where critical privacy discourse also occurs. This limitation highlights the need for a broader and more comprehensive bibliometric analysis that not only quantifies patterns across a longer time horizon, but also evaluates how vulnerable populations, including those defined by intersecting identities, are made visible (or invisible) across the academic ecosystem of privacy research. The present study addresses this gap by analysing a 20-year period (2005–2025) and leveraging bibliometric techniques to offer a multidimensional view of the scholarly attention (or neglect) towards marginalised groups in privacy discourse.

This study adopts a bibliometric analysis approach to systematically map and quantify scholarly literature on privacy research concerning vulnerable populations over a 20-year span (2005–2025). Bibliometric analysis offers a robust quantitative framework to investigate publication trends, thematic evolution and patterns of academic collaboration within a research domain (Passas 2024:1014–1025). This method is particularly well-suited for identifying knowledge gaps, emergent topics and underrepresented communities across large datasets of peer-reviewed publications (Donthu et al. 2021:285).

Data were retrieved from two leading academic databases known for their interdisciplinary reach and high indexing quality: Scopus and Web of Science. These platforms were selected because of their extensive coverage of peer-reviewed journal articles and conference proceedings, which are the primary vehicles of scholarly communication in the field. A structured search query was formulated to capture publications that address privacy and data protection in relation to vulnerable populations. The query combined core privacy terms such as ‘privacy’, ‘data protection’ and ‘information privacy’ with descriptors of at-risk groups. The full query is shown below. The search query was adapted to the syntax requirements of each database. For Scopus, the search was conducted using the TITLE-ABS-KEY field as follows:

TITLE-ABS-KEY ((‘privacy’ OR ‘data protection’ OR ‘information privacy’) AND (‘vulnerable population*’ OR ‘marginalised group*’ OR ‘at-risk group*’ OR ‘underserved population*’ OR ‘disadvantaged group*’ OR ‘high-risk group*’ OR ‘susceptible population*’ OR ‘children’ OR ‘older persons’ OR ‘elderly’ OR ‘refugees’ OR ‘migrants’ OR ‘disabled persons’ OR ‘persons with disabilities’ OR ‘chronic illness’ OR ‘mental illness’ OR ‘physical disability*’ OR ‘pregnant*’ OR ‘drug user*’ OR ‘economically disadvantaged’ OR ‘homeless’ OR ‘domestic violence’ OR ‘incarcerated individuals’ OR ‘transgender’ OR ‘ethnic minority*’ OR ‘religious minority*’ OR ‘minority community*’))

For Web of Science, the search was conducted using the Topic field (TS), which includes title, abstract and keywords:

TS = ((‘privacy’ OR ‘data protection’ OR ‘information privacy’) AND (‘vulnerable population*’ OR ‘marginalised group*’ OR ‘at-risk group*’ OR ‘underserved population*’ OR ‘disadvantaged group*’ OR ‘high-risk group*’ OR ‘susceptible population*’ OR ‘children’ OR ‘older persons’ OR ‘elderly’ OR ‘refugees’ OR ‘migrants’ OR ‘disabled persons’ OR ‘persons with disabilities’ OR ‘chronic illness’ OR ‘mental illness’ OR ‘physical disability*’ OR ‘pregnant*’ OR ‘drug user*’ OR ‘economically disadvantaged’ OR ‘homeless’ OR ‘domestic violence’ OR ‘incarcerated individuals’ OR ‘transgender’ OR ‘ethnic minority*’ OR ‘religious minority*’ OR ‘minority community*’))

Boolean operators and truncation symbols were used to improve sensitivity while maintaining topical relevance. The search was restricted to English-language publications between 2005 and 2025 and limited to journal articles and conference papers. The Scopus search yielded 4051 articles, and Web of Science yielded 4149, resulting in an initial combined dataset of 8200 records. After the removal of 3440 duplicates, a final dataset of 4760 unique articles was retained for analysis.

Bibliographic records were exported from Scopus and Web of Science in BibTeX format and merged using the mergeDbSources() function from the Bibliometrix R package (Aria & Cuccurullo 2017:962). The exported metadata included fields such as:

Duplicates were identified and removed using combinations of DOIs, titles and author fields.

The final dataset was analysed using the Bibliometrix package in R (Aria & Cuccurullo 2017:962). The analytical procedures were carefully designed to address the four research questions outlined in the introduction.

Research Question 1: Which vulnerable populations have received the most academic attention in privacy-related research?

The study conducted:

Research Question 2: Which vulnerable populations are underrepresented in privacy-related research?

To address this question, the study employed:

Research Question 3: How has research on digital privacy and vulnerability evolved between 2005 and 2025?

The study conducted:

Research Question 4: What are the most influential journals, authors and themes shaping this domain?

The study we applied:

To enhance interpretation, the analysis was supported by a series of visualisations, including publication trend graphs, geographic heatmaps of author affiliations, keyword co-occurrence networks and thematic evolution plots. These tools provided both a macro- and micro-level view of how privacy research concerning vulnerable populations has developed over the last two decades.

Ethical clearance to conduct this study was obtained from the Nelson Mandela University Research Ethics Committee (Ref. No. 1418).

Figure 1 illustrates the evolution of scholarly output on privacy and vulnerable populations between 2005 and 2025. The annual volume of publications shows a steady increase over the two decades, rising from 39 publications in 2005 to over 500 in 2024. This upward trend reflects growing academic interest in the intersection of privacy and social vulnerability. The dataset reveals a compound annual growth rate of 11.71%, highlighting an acceleration in scholarly activity, particularly after 2015. A significant spike in publication output occurred during the COVID-19 pandemic period (2020–2023), suggesting that global crises tend to amplify privacy concerns, especially for at-risk and vulnerable populations. This growth pattern indicates a shift in framing privacy not only as a technological issue, but also increasingly as a social and ethical imperative, particularly in contexts involving marginalised or high-risk communities.

The authorship analysis revealed several scholars with substantial contributions to the literature on privacy and vulnerable populations as shown in Figure 2. Wang Y. was the most prolific author, with 46 publications, followed closely by Zhang Y. with 44. Other leading contributors included Cheng Y. (33), Liu Y. (29) and Liu J. (27).

Figure 3 presents a heatmap of corresponding author affiliations by country. Most research originates from high-income countries, with the United States (2557 publications), United Kingdom (896), Canada (569), Australia (523) and China (489) accounting for the largest shares of the literature. Collectively, these countries represent a dominant share of global scholarship on privacy and vulnerable populations. Other notable contributors include Germany (363), India (326), the Netherlands (272), Italy (237), Denmark (223), Spain (219) and Japan (200). A second tier of contributors, each with fewer than 200 publications, includes Sweden (190), Norway (170), France (161), Belgium (155), Switzerland (136), South Africa (120), Taiwan (105) and Finland (103). Importantly, South Africa is the only African country represented among the top 20 contributors, highlighting the broader underrepresentation of the African continent in this domain. This reflects a persistent geographical imbalance in global privacy research and points to the need for greater investment in context-specific studies on privacy and vulnerability across African nations. The lack of contributions from most of the continent suggests that regional privacy concerns, particularly in relation to digitally marginalised populations, may be overlooked in the current global discourse.

This asymmetry in global research output reinforces ongoing concerns about epistemic injustice, where knowledge production is concentrated in the Global North, potentially neglecting the lived experiences, regulatory challenges and cultural dimensions of privacy in the Global South, especially Africa and Latin America. Addressing this gap is essential for fostering more inclusive, equitable and socially responsive privacy research agendas.

To identify dominant research themes and conceptual patterns, a keyword co-occurrence analysis was performed using author-assigned keywords and indexed keywords. Figure 4 presents a combined visualisation of the top 20 author keywords and top 20 indexed keywords side by side, highlighting the most frequently occurring vulnerable group terms.

The most common keywords across both sets include:

Notably, keywords are reported as they appear in the literature. For instance, ‘disabled persons’ is commonly used, even though more respectful and accurate terminology such as ‘people living with disabilities’ is preferable.

To provide a broad thematic overview of the research landscape, a word cloud was generated based on the combined frequency of keywords extracted from both author keywords and indexed keywords across the dataset. This visual representation emphasises the most frequently occurring terms, with larger words indicating higher frequency (Figure 5). The word cloud offers readers an accessible snapshot of dominant topics and recurring concepts within the literature, capturing prevalent research themes without the need for an exhaustive manual review. Key terms that appear prominently include ‘privacy’, ‘female’, ‘human’, ‘male’, ‘humans’, ‘adult’, ‘aged’, ‘child’, ‘article’ and ‘adolescent’, among others. These reflect significant thematic clusters around demographic groups and core privacy concerns. By presenting the entire spectrum of keywords, this section aims to give a clear and comprehensive picture of the vocabulary shaping the field.

An analysis of source productivity revealed that a small number of journals dominate scholarly output in privacy-related research on vulnerable populations. Table 1 presents the 10 most prolific journals based on the number of articles published between 2005 and 2025. BMJ Open emerged as the most productive outlet with 91 articles, followed by PLOS ONE (65 articles) and the Journal of Medical Internet Research (56 articles). Other leading sources include Sensors (52), BMC Public Health (35), BMC Health Services Research (34) and IEEE Access (34). These journals reflect a strong presence of interdisciplinary and health-focused publications, emphasising the prominence of healthcare, public health and digital surveillance themes within the domain. Notably, many of these journals publish open-access content, suggesting an orientation towards accessibility and broad dissemination of findings in privacy scholarship. The concentration of publications within a handful of sources also signals possible gatekeeping effects or disciplinary clustering, which may shape whose perspectives, and which vulnerable populations receive scholarly attention.

Figure 6, based on both author keywords and indexed keywords, consistently shows that key vulnerable groups rarely appear as focal points in privacy research. Among the 10 least represented author keywords, several received no mentions at all, including drug users, economically disadvantaged, minority communities and poor. Other groups like incarcerated individuals, religious minorities, disabled persons, physical disability and persons with disabilities were mentioned only one to three times. Similarly, the 10 least frequent indexed keywords revealed a comparable pattern. Groups such as economically disadvantaged, incarcerated individuals, minorities, minority communities and religious minorities had zero mentions, while ethnic minorities, LGBT, older persons, indigenous and poor appeared only once or twice.

These findings suggest a persistent gap in privacy scholarship concerning intersectional and socially vulnerable populations. Despite these groups facing heightened risks of surveillance, exploitation or data misuse, they remain largely absent from the dominant discourses shaping digital privacy frameworks. The consistent omission of groups such as the economically disadvantaged, incarcerated individuals, minority communities and people living with disabilities raises concerns about epistemic invisibility where the needs, experiences and rights of certain populations are not just underserved but effectively erased from scholarly and policy debates. This underrepresentation not only limits the inclusivity of academic knowledge production, but may also hinder the development of privacy protections that are sensitive to structural inequalities. Moving forward, there is an urgent need for digital privacy research agendas to intentionally engage with these marginalised communities, ensuring that emerging policies, tools and frameworks reflect the full diversity of social vulnerability.

This study offers a comprehensive bibliometric review of privacy-related research concerning vulnerable populations. It reveals consistent growth in scholarly output over the past two decades, with marked increases after 2015. The analysis further highlights geographic disparities in publication volume, a dominant thematic focus on healthcare and children and the persistent underrepresentation of many marginalised groups such as persons with disabilities, religious minorities and economically disadvantaged populations. South Africa emerges as the only African country among the top 20 contributors, ranking 18th overall.

The increasing volume of publications and citations signals growing academic and societal interest in privacy, especially amid technological expansion and global crises like COVID-19. However, the thematic concentration on biomedical ethics and child protection, while important, reveals a narrow conceptual scope. Many vulnerable groups remain peripheral or invisible within mainstream privacy scholarship. These findings are consistent with previous literature on epistemic injustice, where knowledge production often centres Western priorities, neglecting context-specific issues in the Global South. The notable underrepresentation of African nations, with only South Africa featured in the top 20, reinforces concerns about global research imbalances. This suggests that policy and scholarly frameworks often lack sensitivity to regionally specific privacy risks such as those tied to local cultural norms, resource limitations or sociopolitical contexts. Without inclusive representation, dominant privacy frameworks may inadequately reflect the lived experiences of many at-risk communities. These findings both align with and extend existing research in the field. Similar to prior bibliometric and review studies (e.g. Benedetto & Cucchi 2022; Sannon & Forte 2022), this study confirms the dominance of themes related to healthcare, children and data protection within privacy scholarship. However, a key difference lies in the explicit focus on the representation of vulnerable populations, where this study reveals a more pronounced imbalance in scholarly attention than previously reported. While earlier studies have highlighted thematic concentrations, they have not systematically examined which social groups are rendered visible or invisible within the literature. A particularly notable and somewhat unexpected finding is the near absence of certain highly vulnerable groups such as economically disadvantaged communities, incarcerated individuals and religious minorities in both author and indexed keywords. Given the well-documented exposure of these groups to surveillance and data exploitation, their limited representation suggests a disconnect between real-world vulnerability and academic focus. This gap points to an urgent need for future research to better align scholarly inquiry with the lived experiences of marginalised populations.

A major strength of this study lies in its use of a dual-source dataset combining Scopus and Web of Science records, which enhances the comprehensiveness of the bibliometric review. The mixed-method approach combining trend analysis, keyword co-occurrence and visual mapping enables a nuanced understanding of both research volume and conceptual priorities. However, the study is not without limitations. The analysis relied on author-supplied and indexed keywords, which may not fully capture all relevant vulnerable populations, especially when terminology is inconsistent or outdated (e.g. ‘disabled persons’ vs. ‘persons living with disabilities’). Additionally, while rigorous efforts were made to clean and standardise country names and affiliations, variations in metadata quality may have affected the accuracy of geographic attributions. Finally, the absence of full-text analysis means deeper qualitative insights into how vulnerable populations are discussed were outside the study’s scope.

This study underscores the need for more inclusive privacy research that meaningfully incorporates the concerns of underrepresented and marginalised groups. Future studies should expand beyond biomedical and child-centric paradigms to consider a wider array of vulnerabilities, including disability, indigeneity, incarceration and economic disadvantage. Researchers should also adopt inclusive terminology and frameworks that reflect the lived realities of diverse populations, especially in the Global South. From a policy perspective, funding bodies and academic institutions must support scholarship that decentralises Western epistemologies and amplifies region-specific privacy concerns. For practice, developers and regulators should be urged to co-design digital technologies and policies with input from those communities most at risk of privacy violations.