Health sciences librarians' engagement in open science: a scoping review

Objectives: To identify the engagement of health sciences librarians (HSLs) in open science (OS) through the delivery of library services, support, and programs for researchers. Methods: We performed a scoping review guided by Arksey and O'Malley's framework and Joanna Briggs' Manual for Scoping Reviews. Our search methods consisted of searching five bibliographic databases (MEDLINE, Embase, CINAHL, LISTA, and Web of Science Core Collection), reference harvesting, and targeted website and journal searching. To determine study eligibility, we applied predetermined inclusion and exclusion criteria and reached consensus when there was disagreement. We extracted data in duplicate and performed qualitative analysis to map key themes. Results: We included fifty-four studies. Research methods included descriptive or narrative approaches (76%); surveys, questionnaires, and interviews (15%); or mixed methods (9%). We labeled studies with one or more of FOSTER's six OS themes: open access (54%), open data (43%), open science (24%), open education (6%), open source (6%), and citizen science (6%). Key drivers in OS were scientific integrity and transparency, openness as a guiding principle in research, and funder mandates making research publicly accessible. Conclusions: HSLs play key roles in advancing OS worldwide. Formal studies are needed to assess the impact of HSLs' engagement in OS. HSLs should promote adoption of OS within their research communities and develop strategic plans aligned with institutional partners. HSLs can promote OS by adopting more rigorous and transparent research practices of their own. Future research should examine HSLs' engagement in OS through social justice and equity perspectives.


INTRODUCTION
Open science (OS) is a global movement that promotes transparency and reproducibility in research, contributes to biomedical education and training, and facilitates secondary research [1]. In recent years, health sciences librarians (HSLs) have participated in a range of OS projects, but to date there have been no broader studies identifying this complex and varied support. To address this gap, we conducted a scoping review of HSLs' efforts to support OS researchers and advance OS principles through library-led services and programs.

Definitions and background
The approach, methodology, and analysis of this review was informed by several definitions. OS has been defined as "the practice of making everything in the discovery process fully and openly available, creating transparency, and driving . . . discovery by allowing others to build on existing work. The six . .  [2]. OS also refers to an array of practices promoting openness, integrity, and reproducibility in research [3]. A glossary of terms used in this paper is available on the Open Science Framework (OSF) [4].
Vicente-Saez and others embarked on a systematic review to develop a better definition of OS [5]. After reviewing thirty-six studies, the authors defined OS as "transparent and accessible knowledge that is shared and developed through collaborative networks" [5]. Several

Building library support and rationale
HSLs have played key roles in supporting open scholarship [10,11]. As partners in promoting OS, HSLs have increased their participation in research data management [12], open access (OA) initiatives [13][14][15], and citizen science [16]. Since 2010, the EU and member states have implemented open data policies [17], and librarians have increased their engagement by improving OS governance and e-infrastructure [18]. For HSLs serving biomedical researchers, promoting OA and open data practices is key to achieving an open, inclusive, and equitable science ecosystem [19]. The Medical Library Center of New York was an early advocate of open scholarship and "the FAIR guiding principles for scientific data management and stewardship" (making data findable, accessible, interoperable, reusable) [20].
Our objectives in this review are to assess the scope of HSLs' support of OS, examine strategic approaches they have taken, and identify the impact of library services in OS. We are interested in how HSLs develop OS services, demonstrate their impact over time, and align user services with broader institutional goals and resources via partnerships, curriculum integration, policies, and infrastructure.

METHODS
Our scoping review was guided by Arksey and O'Malley's framework [21] and Joanna Briggs' Manual for Scoping Reviews [22]. We developed a review protocol to guide the process and shared it via the OSF in March 2020 [4]. For reporting, we used PRISMA-S for our searches [23] and PRISMA-ScR for the review itself [24].
The following research questions (RQ) guided our review: In our review, the term "researcher" refers to any scientist, clinician or practitioner, student, or other individual working in a hospital or health care setting, biomedical lab, government agency, or academic or research institution who is engaged in research or in the process of learning how to conduct research.

Data sources and literature searching
Scoping reviews employ broad exploratory searches of relevant studies to determine key characteristics of a subject [25]. We searched five bibliographic databases from 2010 to 2020: MEDLINE (Ovid); Embase (Ovid); CINAHL (EBSCO); Library, Information Science and Technology Abstracts (LISTA) (EBSCO); and the Web of Science Core Collection (Science Citation Index Expanded, Social Sciences Citation Index, Arts and Humanities Citation Index, Conference Proceedings Citation Index-Science, Conference Proceedings Citation Index-Social Sciences and Humanities, Emerging Sources Citation Index).
We limited our searching for the ten-year period from 2010 to 2020 because OS, as an umbrella term of open practices beyond open access, was not widely discussed before 2010. Although some public access policies, such as the National Institutes of Health (NIH) Public Access Policy, were established before 2010, the first studies examining their application in HSLs began to appear in 2010. In our early scoping searches, we also found a high number of relevant studies published after 2015.
In addition to database searching, we searched seven journals in health sciences librarianship and three association websites. We performed reference harvesting and citation searching in Web of Science and Google Scholar using a file of highly cited peer-reviewed and grey literature found in our searches (Supplemental file S1). Searches were conducted in March 2020 and updated monthly until December 2020.

Search strategy development
Our search terms were developed based on our RQs. We grouped them into three concept blocks: 1) librarians, libraries, and variations of "health sciences librarians or libraries"; 2) research trends in OS inclusive of our definitions, pillars, practices, principles, tools, and platforms [2,3,[5][6][7][8]; 3) HSLs' actions, roles, and support and outcomes of providing library services and programs in OS (Supplemental file S2).
We devised our search sets in MEDLINE using two main concept blocks (health sciences librarians and OS) and tested the third group of concepts. We used the third concept block in LISTA only as it reduced search sensitivity too much in the other databases. We imposed no limits by language or study type but limited to studies published from 2010 to 2020.
Searches were performed by two authors (DG, KR) and peer reviewed by two other authors (LF, MR) using the Peer Review of Electronic Search Strategies checklist [26]. We created strategies in MEDLINE and Embase on Ovid and adapted them to the other databases and platforms. We include our MEDLINE search strategy (Supplemental file S3) and complete details of the other searches via the OSF platform [4].
Key journals, association websites, meetings, and abstract searching

Environmental scanning
We performed an environmental scan of OS-related library activities, workshops, and conferences as reported on library and university web pages, subject guides, and association websites in North America, Europe, and Australia [4]. We also performed early exploratory searches for perspectives from Latin America in LILACS and SciELO and from continental Africa in the WHO Global Index Medicus. Our environmental scan helped to identify an overall search strategy as well as search terms, key concepts, and sources of evidence most relevant to the review.

Citation management
We used Zotero, an open source tool, for citation management. Our searches yielded 8,173 studies from database, journal, and conference searching inclusive of reference harvesting and citation tracking [4]. The PRISMA-ScR flow diagram is shown in Figure 1. We performed deduplication in Zotero and imported unique citations into Covidence for screening.

Our open practices as researchers
As a team of researchers, we are committed to OS principles and make our literature searches and datasets open, transparent, and publicly available. Our methodology aims to adhere to open principles as much as possible, which was increasingly important given that we conducted the study during COVID-19. However, although we used open-source alternatives when and where possible, we relied on library subscriptions, proprietary products, and platforms to perform the bulk of the review.

Inclusion and exclusion criteria
We considered all sources to allow for inclusion of any and all types of studies (including peer-reviewed and nonpeer-reviewed studies) and methodologies (e.g., case reports, mixed methods, qualitative studies, quantitative studies). We included studies that were written in or could be translated into English.
Studies were included when they discussed the following topics: Studies and other literature were excluded if they were published before 2010 or did not discuss or describe: • OS library services, leadership, or support in any way relevant to our research questions; • OS-related library services aligned with our definitions of OS (e.g., discussed data management training but not for improved transparency or reproducibility); • Digitization or accessibility initiatives as an OSrelated library service; or • Involvement of HSLs or health sciences libraries.

Study selection and screening
We used a two-step process for screening, first considering titles and abstracts only and then moving to full-text review.

Title and abstract screening
After deduplication in Zotero, we moved the citations to Covidence for screening. Covidence found a further ninety-three duplicates, which were removed. We independently coded and assessed the titles and abstracts of retrieved citations in duplicate using the predetermined selection criteria. Reviewers were not blinded to author or journal names. When a title or abstract was not in English, we included it and ordered translations using one of our review member's institutional translation services. After basic screening was completed, we met to discuss references selected for inclusion. We resolved our disagreements through discussion and majority vote.

Full-text screening
We identified 319 studies for full-text review. We downloaded the documents into Covidence from OA journals and repositories and by accessing our institutional journal subscriptions and interlibrary loan services. For full-text screening, we adopted a process similar to that employed during title and abstract screening. Pilot testing involved assigning a random set of studies to each reviewer and then discussing and improving the process as a team. We resolved disagreements between reviewers during our regular meetings and tracked our decisions to include or exclude in Covidence.

Data extraction
Data extraction for a scoping review is referred to as "data charting" [28]. We tested a data extraction form in Google Sheets and made changes based on this testing. Each reviewer was then assigned a second larger sample set of studies to review. We iteratively improved the form to include the following data fields: citation, author country, publication type, language, study aims, study methods, population, setting, drivers, service type, OS category, service impact, institutional integration, recommendations, and notes (Supplemental file S4). We extracted information from full-text publications and materials by taking into account each variable.

Thematic analysis and categorization
We analyzed the studies using a method by Braun and Clarke [29]. Following the step-by-step guidelines in their method (familiarization with the data; generation of initial coding; searching, reviewing, and defining themes [29]), we identified and highlighted certain themes and patterns across the studies. To focus on identifying, analyzing, and reporting patterns within the data, the work of Thomas and Harden informed our thematic and narrative synthesis [30]. In analyzing each paper, we noted emergent themes based on study participants, types of library services and programs provided, and study outcomes. We coded the themes and placed them into categories using FOSTER's taxonomy in order to apply a consistent, structured set of terms. The taxonomy provides an overview of core topics in OS and is used by FOSTER to organize teaching and learning materials on their website [6]. To classify publications, we were guided by publication types as defined by the Medical Subject Headings thesaurus [31].

RESULTS
We identified fifty-four studies that met our inclusion criteria, a summary of which is available in  Table 2).
Most of our studies fit the definition of a case study, "a descriptive and exploratory analysis of a particular library, service or event." We adopted the definition of research articles as "the predominant publication type for articles and other items indexed for NLM databases" [4,31] unless one of the other categories was more appropriate.
Studies were published in 27 interdisciplinary scholarly and journal sources. Several studies appeared in the Journal of the Medical Library Association (n=8, 15%) and the GMS Medizin-Bibliothek-Information (n=7, 13%). Bibliographic databases containing the most studies were LISTA (n=22, 41%), Web of Science (n=12, 22%), CINAHL (n=8, 15%), and MEDLINE (n=7, 13%). We include a list of 54 studies and journal sources, traced back to their original database, in Supplemental file S5.
Our thematic analysis reveals the multidimensional nature of HSLs' support of open science. We clustered findings into three themes based on our research questions: 1) key actions, barriers, and drivers of OS in HSLs; 2) types of OS services described; and 3) roles and stakeholders involved in providing HSL-specific support in OS ( Figure 4). The relevant subthemes are clustered below each main theme, and the number of studies addressing each subtheme is indicated in brackets.     [33]; helping researchers to comply with funder mandates such as the NIH Public Access Policy, EU Open Science Policy, and NIH Rigor and Reproducibility requirements (n=11, 20%); undertaking OS initiatives at the campus level, for example designing library services to align with institutional drivers such as university policies, working with faculty members and instructors to teach OS and its principles, and developing OS-specific infrastructure (n=8, 15%) or cross-institutional projects to initiate or improve OA (n=4, 7%); and initiating OS services because other libraries were doing so (n=3, 6%). Specific barriers to providing OS services were not explicitly mentioned.

Theme 2: HSLs' OS services and support Advocacy and outreach for OS services
HSLs engage in advocacy and outreach for OS in many ways, with most examples coming from academic institutions, including advocating for OA and OS in the academic health context, including supporting and promoting OA funds (n=7, 13%); showcasing OA and OS research efforts on campus by making presentations, holding workshops, and pursuing cross-institutional initiatives (n=4, 7%); coordinating various stakeholders on campus to support the adoption of open publishing models as well as open educational resources (n=4, 7%); serving as mediators and experts in local open data efforts in academic institutions or being members in national organizations (n=2, 4%); and promoting the principles of citizen science by building connections between researchers and local communities, scheduling citizen outreach, and codeveloping community-researcher-based networks (n=2, 4%). We found no evidence of advocacy or outreach for open source-related services.

Open access services and resources
HSLs help to promote OA to researcher communities by offering the following services and resources: raising awareness of OA principles and their adoption (n=8, 18%); developing infrastructure to assist in depositing OA publications for policy compliance (n=5, 9%); managing institutional OA publications (n=2, 4%); developing institutional repositories (n=5, 9%) and digital archives (n=2, 4%); creating and maintaining library guides to disseminate key library services for OA (n=5, 9%); and encouraging compliance with national funder policies, specifically the NIH Public Access policy and EU Open Access policy (n=4, 7%).

Open data services and resources
HSLs help to promote open data by offering the following services and resources: providing support for the reuse and sharing of open data (n=6, 11%); offering digital object identifier (DOI) services for datasets (n=2, 4%); and providing technical and advisory support for the Open Science Framework (n=2, 4%).
HSLs have developed the following resources to support open data: repositories to store and share institutional research data (n=6, 11%); data catalogs to improve the discovery of institutional research data that cannot be publicly shared (n=2, 4%); and open data LibGuides (n=2, 4%).

Open science resources and services
HSLs help to promote the broader theme of OS by offering the following services and resources: coordinating and organizing OS conferences, meetups, and events (n=3, 6%); and supporting campus OS policies (n=2, 4%).
In addition, HSLs provide OS instructional and training services on the following topics: research reproducibility, transparency, and OS best practices and principles (n=4, 7%); data science training, specifically the use of R and Python (n=2, 4%); OS-specific tools, specifically research protocol sharing software (n=1, 2%); and adoption, use, and attribution of standards used in research projects (n=1, 2%).
HSLs provide OS support on institutional and grantfunded research projects including: collaborating on initiatives to improve research workflow infrastructure and guide development and sharing of research outputs (n=3, 6%); and developing infrastructure to support data citation and the implementation of data standards used for research projects (n=3, 6%).

Citizen science services and resources
HSLs help to promote citizen science by offering the following services and resources: providing access to tools and technology to help community members better understand science and ongoing research projects at the librarian's institution (n=1, 2%); developing a citizen science toolkit for researchers and community members (n=1, 2%); training citizens on institutionally developed research projects where citizens then vote on which projects receive grant funding (n=1, 2%); and teaching the value of citizen science (n=2, 4%), principles of information literacy (n=1, 2%), and new methods of data collection (n=1, 2%).

Theme 3: Roles and stakeholders involved in providing HSL OS services and support
In addition to their teaching and consultation roles in OS, HSLs assume several key and prominent roles within their institutions by offering the following services and support: advocating for and compliance of OS throughout the research process (n=20, 37%); collaborating on institutional OS-related policy development, specifically policy as it relates to OA and open data (n=6, 11%); building community within institutions with respect to OS principles and, in Europe, establishing connections with ongoing OS national initiatives (n=4, 7%); and providing metadata support (n=1, 2%) and digital preservation (n=1, 2%) for OA and open data initiatives in academic institutions.
While the types of roles articulated across the included studies were described broadly, there was no mention of specific job titles or formal institutional responsibilities that were established to carry out these roles.

DISCUSSION
This scoping review identified fifty-four studies with authors from eleven countries and three continents who worked mostly as government and academic HSLs. The published literature has increased considerably since 2010, with 52% of articles published since 2018, signaling the topic's increased importance for HSLs. We extracted a total of fifty-seven unique sub-themes, highlighting a range of OS services provided by HSLs. Based on our analysis, we identified the following areas that could benefit from further exploration: 1) geographic concentration of OS support by country and continent, 2) social justice issues related to OS, 3) OS-specific gaps to address in future research, and 4) the lack of evidence to support the success of HSL OS services.

Geographic concentration of OS support
Our review highlights trends and variations in how HSLs from different countries implement OS services. Studies published in European countries, for example, revealed strong coordination between national efforts, such as the European OA policy and release of Plan S for OA publishing, and adoption of these efforts at the institutional level. Furthermore, HSLs coordinated activities such as working with their governments to identify and implement OS services; contributing to policy development at an institutional level and informed by national policy; and, in collaboration with national partners, developing infrastructure (e.g., institutional repositories) and resources (e.g., OA reporting for policy compliance). Two studies from continental Africa (Nigeria and Uganda) demonstrated strong alignment with national initiatives. The Uganda study in particular outlined a project that brought together students and researchers at a government-sponsored event to explore open education resources.
By comparison, North American studies predominantly published by US institutions demonstrated that HSLs use national policies (e.g., NIH public access policy) to justify their development of OS services, but these services are not typically created in direct coordination with national governments and stakeholders. North American HSLs may find it useful to look at examples in Europe to determine how OS services could be developed alongside national partners. Further, given the shortage of perspectives from regions in Oceania, Latin America, and continental Africa, HSLs from these jurisdictions should be encouraged to publish their research more generally and also to highlight best practices that are emerging in creating those important partnerships.

Open science and social justice
Only one of the fifty-four studies mentioned OS's potential to advance equity and social justice perspectives jmla.mlanet.org [88]. Given that one of the goals of OS is to make research results more widely available, we were surprised to learn that equity and social justice were not mentioned more broadly as drivers of OS services. Many biomedical researchers and clinicians are increasingly invested in anti-racist work, so there is an opportunity to highlight OS practices as an alternative to the traditionally closed, white supremacist models of scholarly publishing [89]; however, there is some concern that OS also suffers from similar exclusive practices [90]. HSLs with institutional diversity, equity, and inclusion (DEI) initiatives might consider aligning OS services in support of those programs by advocating for DEI funds to support OA publishing fees for early career and underrepresented researchers. Given the evidence that they are more likely to be excluded in OS, there should be greater focus on supporting women, minority groups, and Black, Indigenous, and People of Color (BIPOC) [89]. Another important consideration is to make more explicit the connections between OS and equity in DEI programming by providing opportunities to groups historically marginalized or excluded from open research due to disparities in power and privilege [91].

OS-specific gaps in the literature
Our review highlights numerous gaps in the HSL OSspecific literature. First, while there was a high frequency of HSL OS studies in academic settings, we found no studies in clinical or hospital settings. While this may reflect the often smaller budgets and limited staff time at these institutions, we argue that OS support is especially important for under-resourced institutions, where researchers and clinicians often publish in journals they cannot afford to license and would benefit from more awareness of OA models. We recommend that academic HSLs partner with hospital librarians in their area to collaborate on OS services to support mutual goals. We suggest that citizen science and OA might be good places to start.
Second, only a small number of studies discussed providing targeted OS support to specific groups (e.g., nursing, basic science), while the remainder focused on broad, institution-wide services. We recommend that future research focus on collaborating, incorporating, aligning, and evaluating HSL OS services with user communities that have a logical connection to OS, such as population and public health. Finally, our review highlights a lack of publications in citizen science and its potential to further patient-centered research [92]. Despite the two examples we found, there were few examples of citizen-specific or community-based partnerships. As HSLs continue to provide library services and promote OS principles, we believe that more concerted efforts are needed to engage local communities and individuals (e.g., patients, research participants, neighborhood organizations) in projects at an institutional level. A shift toward more support for citizen science would also demonstrate HSLs' commitment to advocating for OS beyond their institutions and help to connect researchers with community partners and patient advocacy groups to ensure research results are shared to maximize equity and impact.

Lack of evidence of the success of HSL OS services
Our review highlights the limitations in the HSL literature related to evidence-based OS practices. While the included studies describe how HSLs are providing OS services, more empirical research is needed to demonstrate the impact of these services and the costs versus benefits of these services over time, including rigorous assessment and sustainability of library service models. Due to a lack of robust study designs, we could not assess OS services provided by HSLs or determine for whom the OS services were intended beyond general mention of students and faculty. Interestingly, many studies stated in their recommendations the need for additional support from their institutions to advance OS and its principles. Empirical evidence gathered from HSL OS services could help influence institutional support for additional HSL resources and personnel.

Limitations
This scoping review has some limitations. By limiting our searches to a finite set of bibliographic databases and websites, we acknowledge that we may have excluded important perspectives of HSL OS services, particularly from continental Africa and South America. In addition, by limiting the date range of our searches from 2010 to 2020, we may have missed studies published before 2010. We opted to limit our searching to the previous ten years to ensure that the body of literature would address OS services more broadly, rather than emphasize open access, which was the primary OS service described before 2010. To account for any relevant OA studies published before 2010, we quote from and cite those papers in our introduction.

Conclusion
HSLs are engaged in providing various OS library services and programs and play prominent roles in advancing OS practices and principles worldwide. However, more formal studies should be conducted to assess their support of OS researchers and to determine the best and most sustainable library services and programs going forward. Future research should focus on researchers' needs in OS and robust assessment of library service models specifically designed to meet them. Furthermore, HSLs can amplify OS by adopting and promoting more rigorous and transparent research practices of their own. Future research should also examine HSLs' engagement in OS through equity and social justice perspectives.