Compliance with the first funder open access policy in Australia

Introduction

The National Health and Medical Research Council is Australia’s leading funder of medical research. Aligning with the global implementation of open access policies by funding bodies, governments, and institutions, the Council implemented Australia’s first national-level open access policy commencing 1 July 2012. Founded on the Budapest Open Access Initiative (2002), the Council’s definition of scholarly open access is the ‘availability of research outputs via the internet, such that any user can find, freely access … and otherwise use and reuse … [with] an appropriate licence’ (National Health and Medical Research Council, 2018a, p. 5).

This article reports on a research project that examined compliance with the Council’s Open Access Policy during the first two full years (2013–2014) of the Policy’s introduction (Kirkman, 2018). The study identified Council-funded articles in Web of Science and applied bibliometric methods to assess the extent to which funded publications met the requirements of the Policy. As the first such study in Australia, this research establishes a benchmark for future studies and highlights potential issues in the Australian open access environment.

Background

A vast body of literature surrounds scholarly open access, although only a small proportion focusses on compliance with funder policies. The theoretical foundation for funder mandates emanates from the Budapest Open Access Initiative (2002), the first formal statement relating to open access and research literature. Notwithstanding, an examination of the Registry of Open Access Repository Mandates and Policies demonstrates that policies differ in their requirements. Some policies mandate immediate and published open access, while others specify the deposit of accepted manuscripts in repositories.

The United States National Institutes of Health Public Access Policy requires all funded peer-reviewed articles to be publicly accessible by publishing in participating publishers’ journals or depositing author manuscripts in PubMed Central, an archive maintained by the National Library of Medicine (US. Dept. of Health and Human Services, 2008). In Canada, the Tri-Agency Open Access Policy on Publications mandates that journal articles arising from agency-supported research be freely accessible within 12 months of publication. For almost a decade, PubMed Central Canada functioned as the repository for author manuscripts funded by the Canadian Institutes of Health Research. Low deposits led to the decommissioning of PubMed Central Canada in early 2018 and the transfer of manuscripts to the National Research Council digital repository with authors required to deposit in institutional repositories (Canadian Institutes of Health Research, 2017).

In Europe, Plan S is an initiative of cOAlition S and Science Europe comprising a group of research funding organisations supported by the European Commission and the European Research Council. The fundamental principle of Plan S is immediate open access with Creative Commons Attribution licensing (cOAlition S, 2019a, 2019b). Plan S embraces funder support for article processing charges to publish in fully open access journals listed in the Directory of Open Access Journals but not for publishing open access in hybrid subscription journals. According to cOAlition S, hybrid subscription journals lack success in the delivery of immediate open access at reasonable costs (cOAlition S, 2019c). Delayed open access, a model used by learned societies to provide access to their journals by members but with embargoes in place for non-members, is not part of Plan S, although many medical societies provide global health research free of charge (Haug, 2019). Plan S supports the deposit of manuscripts in repositories at the time of acceptance along with high-quality article metadata incorporating grant information. A new generation of repository platforms will be necessary to fulfil the technical requirements of systems under Plan S (cOAlition S, 2019a, 2019c).

The Wellcome Trust, an independent global charitable foundation that funds medical research, revised its open access policy to align with the Plan S principle of immediate open access. The Trust endorses the use of its grants for article processing charges to publish in full open access journals and those journal-titles with agreements with the National Library of Medicine to deposit open access versions in PubMed Central or Europe PubMed Central. Openly accessible funded research articles published in partial open access journals are also acceptable under the Wellcome Trust mandate (Wellcome Trust, 2020a, 2020b). The Trust’s publishing platform, Wellcome Open Research, supports immediate open access by Wellcome-funded researchers.

The United Kingdom Research and Innovation national funding agency includes several research councils, Innovate United Kingdom, and Research England. Research England manages the Research Excellence Framework on behalf of higher education funding bodies in the United Kingdom. The Research Excellence Framework 2021 policy requires open access in repositories for all articles and papers from journals and conference proceedings with International Standard Serial Numbers (Research Excellence Framework, 2019a, 2019b). A higher education review into institutions’ progress in delivering funders’ open access policies uncovered the demanding work undertaken by staff of institutional repositories, the complexities of publisher policies, and the lack of publisher and funder policy alignment (Research England, et al., 2018). Despite the higher workloads and tighter time-frames required to meet the goals of the Research Excellence Framework 2021, institutional repositories remain inadequately resourced (Herrmannova, et al., 2019; Ten Holter, 2020). In February 2020, United Kingdom Research and Innovation launched a review proposing a single open access policy and recommending immediate availability of funded research articles in journals, publishing platforms, institutional or subject repositories (United Kingdom Research and Innovation, 2020).

The chief principle behind the open access policies of Australia’s leading funders, the Australian Research Council and the National Health and Medical Research Council, is the wide dissemination of publicly funded research (Australian Research Council, 2017; National Health and Medical Research Council, 2012, 2014a, 2014b, 2014c, 2018a, 2018b, 2018c). Both policies require the metadata of funded articles, along with grant information, in at least one institutional repository as soon as possible after publication. The main difference between the National Health and Medical Research Council and the Australian Research Council policies is that the latter includes scholarly books, book chapters, as well as non-traditional research outputs. The other difference relates to timing. The Australian Research Council policy relates to all funded research outputs since 1 January 2013. The National Health and Medical Research Council policy applies to journal articles (and later peer-reviewed conference papers) published from 1 July 2012, regardless of the date of the grants that supported the research. Australia’s research assessment body, Excellence in Research Australia, requires a statement as to whether a submitted output is openly accessible, but open access is not part of the evaluation process (Australian Research Council, 2016).

This study of the National Health and Medical Research Council’s Policy commenced when the 2012 version was in force and still in progress with the release of later versions and updates (see Table 1). This paper uses the uniform title Open Access Policy to refer to all versions of the Council’s Policy and to accommodate title variations. All policy versions retain the Council’s rationale for maximising the benefits of publicly funded research through open access to ensure the dissemination of funded publications to researchers and the broader community. Journal articles are the primary focus of all policies since 2012, and this study reflects that same focus. The Council’s most recent policy continues to require open access within twelve months of publication. A vital feature of the Open Access Policy is the requirement to deposit the metadata, including the grant identification number(s), of all funded articles into institutional repositories, with the chief investigator of the grant being responsible for the deposit.

The Council’s recommendation to grant holders is to select journals that align with the Findable, Accessible, Interoperable, Reusable (FAIR) principles and have Creative Commons Attribution licensing (National Health and Medical Research Council, 2018b). Journals that provide immediate open access in fully open access and hybrid subscription journals are automatically compliant with the Council’s Policy. The use of grants to pay for article processing charges for immediate open access is acceptable for approved projects. A small number of journals provide free access without publishing fees, although some titles lack Creative Commons licensing. Compatible with the Council’s Policy are journals with embargoes on open access of 12 months or less. When selecting journals, Council grant recipients face a range of journal open access models.

Publishing in journals listed in the Directory of Open Access Journals is known as full open access (Taubert, et al., 2019) but also as gold open access (Piwowar, et al., 2018). Some studies use the term gold to describe all open access in journals (Larivière and Sugimoto, 2018), while other investigations associate gold with immediate open access journals (Jubb, et al., 2017). Hybrid open access in subscription journals is another model used by researchers (Borrego, 2016; Piwowar, et al., 2018). Bronze open access refers to journals that provide public access without re-use licensing, such as gratis open access journals (Piwowar, et al., 2018). However, gratis and partial open access are also terms used to describe specific journal models (Martín-Martín, et al., 2018; McVeigh and Pringle, 2005; Suber, 2012; Willinsky, 2006). Many researchers view delayed open access as a distinct category (Borrego, 2016; Laakso and Björk, 2013; Laakso and Lindman, 2016; Taubert, et al., 2019), but Piwowar, et al. (2018) found that many bronze articles were from journals with embargoes. As the comparison of studies on open access requires consistent terminology, this study uses a typology based on journal open access models outlined in Table 2. The only open access option for authors who publish in subscription-only journals is self-archiving in repositories or on author websites.

The deposit of author accepted manuscripts in institutional or subject repositories is compliant with the Council’s Open Access Policy. Australian institutional repositories are the primary locations, with PubMed Central also recognised. The Council identified the National Library of Australia’s portal, Trove, as the central platform for locating its funded publications in institutional repositories (National Health and Medical Research Council, 2015; 2019). Critical to the understanding of open access in repositories is the differentiation between the three main versions described in Table 3. The mandated version under the Council’s Policy is the author accepted manuscript, the version of the article as accepted by the journal after peer review and revisions (National Health…, 2014a, 2014b, 2014c, 2018a, 2018b, 2018c). Most publishers have policies specifying their conditions for the deposit of author accepted manuscripts in repositories, including the attachment of statements concerning version and licensing, as well as embargoes that regulate the timing of the deposit. Author manuscripts in PubMed Central are identifiable by a banner verifying the compliance of the version with the National Institutes of Health Public Access Policy or the policies of partner organisations. The deposit of publishers’ versions in repositories is not a requirement of the Council’s Policy, although encouraged; preprints are non-compliant (National Health and Medical Research Council, 2014a, 2014b, 2014c, 2018a, 2018b, 2018c).

As noted previously, this is the first study of a funder open access policy in Australia. In Spain, the investigation by Borrego (2016) was the earliest study to provide a comprehensive measurement of open access compliance with a research funder. Bakker et al. (2017) examined the mandate of the Multiple Sclerosis Society of Canada, and Larivière and Sugimoto (2018) undertook a large-scale analysis of compliance with the policies of 12 funding agencies in Canada, Europe, the United Kingdom and the United States. Sources used in studies to verify open access articles include Google Scholar, subject repositories such as PubMed Central, and institutional repositories (Borrego, 2016; Bakker, et al., 2017). The Unpaywall database also links users directly to open access content from journals listed in the Directory of Open Access Journals, hybrid subscription journals, as well as PubMed Central and institutional repositories (Piwowar, et al., 2018). Using Unpaywall, Larivière and Sugimoto (2018) identified open access articles in journals (gold), repositories (green), and those that were both green and gold.

Table 4 provides a comparison of the findings of select studies into open access, including funder investigations. Larivière and Sugimoto (2018) revealed rates of open access compliance under funder policies that ranged from 26.8% for the Social Sciences and Humanities Research Council of Canada to between 86.6% and 91.4% for publications funded by the Wellcome Trust and the National Institutes of Health. Under the Spanish Government policy, Borrego (2016) found total open access of 45.6% comprising 23.8% in journals and 21.8% in repositories. Jubb, et al. (2017) reported that 37% of research outputs in the United Kingdom were immediately open access, with this percentage increasing to 53% after 12 months, mostly due to the contribution of delayed open journals and the deposit of open access versions in repositories. Most studies in Table 4 uncovered low levels of open access in hybrid subscription journals. Jubb, et al. (2017) proved the exception by demonstrating the growth of hybrid open access from a low base of 2.7% in 2012 to 15.4% in 2016, which they attributed to United Kingdom funder policies.

Open access versions in repositories are crucial components of open access policies. Larivière and Sugimoto (2018) reported the high level of open access versions in PubMed Central (43.3%) under the National Institutes of Health Public Access Policy, along with the large proportion of articles that were both open access in journals and repositories especially under the Wellcome Trust mandate. Borrego (2016) recorded a level of open access in repositories of 21.8%, most of which was in subject repositories (15%). International and Australian studies described the low deposit of open access versions in institutional repositories, owing to authors’ preference for subject repositories, negative attitudes towards institutional repositories, and the lack of knowledge of repository procedures (Bakker, et al., 2017; Björk, et al., 2014; Borrego, 2016; 2017; Henty, 2014; Huggard, et al., 2017; Kruesi, et al., 2019a; Kruesi, et al., 2019b; Laakso, 2014; Melero, et al., 2017; Swan and Brown, 2005; Zhu, 2017). Another significant barrier to open access in repositories is the length of some publishers’ embargoes on the deposit of author accepted manuscripts (Gray, 2018; Rumsey, 2017). For example, Khoo and Lay (2018) found that Elsevier’s embargoes of more than 12 months affected compliance with funder open access policies.

The Council’s Open Access Policy also requires the deposit of the metadata of all funded articles in Australian institutional repositories. The inclusion of funder and grant identification number(s) in the metadata is the responsibility of the grant’s chief investigator (National Health and Medical Research Council 2014a, 2014b, 2014c, 2018a, 2018b, 2018c). Sources of metadata exist in online repositories such as PubMed Central, funder records, and databases such as Scopus, Web of Science, and PubMed, but researchers found that acknowledgment of funders is not universal (Álvarez-Bornstein, et al., 2017; Borrego, 2016; Grassano, et al., 2017; Wang, et al., 2015; Wang, et al., 2012).

Although the literature demonstrates an increase in the range of journal open access models, it also records the low deposit of author accepted manuscripts in institutional repositories. The Council’s Open Access Policy mandates open access and the deposit of metadata in institutional repositories for all funded journal articles, however, to date there is no understanding of the extent of compliance with the Policy and the open access publishing choices made by grant recipients. This study aimed at a greater understanding of the Council’s Open Access Policy during its formative period (2013–2014) by identifying:

• the overall extent of open access compliance of Council-funded articles under the Policy;
• the proportion of open access Council-funded articles in journals;
• the proportion of Council-funded articles with author accepted manuscripts in Australian institutional repositories (and not open access in journals);
• the extent of metadata for Council-funded articles in Australian institutional repositories; and
• the journal open access models and author accepted manuscript policies of the journal titles in which Council grant recipients published.

Methods

The study applied bibliometric research approaches to gather and analyse the data. The addition of funding acknowledgement fields to Web of Science in 2008 and Scopus in 2013 created greater possibilities for bibliometric analyses facilitated by the automatic extraction of data (Álvarez-Bornstein, et al., 2017). Studies identified Web of Science as having more funder data compared to other databases (Kokol and Vošner, 2018; Paul-Hus, et al., 2017). The funding acknowledgement coverage for the years 2013 and 2014 was the main reason for the selection of Web of Science as the source database for the population of this study.

Web of Science has three funding acknowledgement search fields: these being funding text, funding agency, and funding grant. Some researchers found that the funding text field, based on the acknowledgement in the original publication, returned a higher number of records than the funding agency field (Paul-Hus, et al., 2016; Tang, et al., 2017). Based on previous studies, this study used both the funding agency and funding text search fields to identify the article population. Acknowledged are issues with the indexing criteria used by Web of Science including the exclusion of journals and articles for which the funding information is in a language other than English, and the lack of standardised funding agency identifiers (Álvarez-Bornstein, et al., 2017; Begum and Lewison, 2017; Costas and van Leeuwen, 2012; Larivière and Sugimoto, 2018; Mongeon and Paul-Hus, 2016; Paul-Hus, et al., 2017; Tang, et al., 2017; Wang, et al., 2015; Wang, et al., 2012). Wang and Shapira (2011) highlighted the importance of data cleaning and the creation of a thesaurus of funding body variations.

For this study, the refine results panel of Web of Science proved useful in creating a list of names and acronyms for the National Health and Medical Research Council. Two separate advanced search strategies identified Council name variations in the funding agency and funding text fields, with the searches combined and refined by the respective publication year (2013 and 2014). The results exported to Microsoft Excel became the master-spreadsheet after the removal of duplicates, non-journal publications, and articles in which the only reference to the Council was to its research ethics codes. The population comprised 12,753 Council funded journal articles published in 2013 and 2014. Limited time and resources for the study prompted the decision to analyse a random sample from the full set of publications. The sampling method, illustrated in Figure 1, determined the sample that represented 25% of the total population and comprised 3,190 articles.

Data collection

The 3,190 articles in the sample provided the primary data and formed the data collection dataset (a spreadsheet) to determine compliance with the Council’s Open Access Policy. The dataset included coding columns for data on the extent of open access articles in journals, as well as author accepted manuscripts and metadata in institutional repositories. Numerical coding ensured uniformity of data entry, necessary for later data analysis of the dataset. Data collection took place in January 2016 and January 2017, both periods being 24 months after the last possible date of publication for articles published in 2013 and 2014. This extended time frame did not affect data collection on immediately published open access articles but had the potential to produce a more generous account of deposits in institutional repositories.

All searching occurred off-campus to ensure that the results excluded subscription-only access. Google Scholar was the initial source for locating links to publishers’ websites. The publisher’s link either connected to the final published article or a payment request confirming that the publication was behind a paywall. For articles not open access in journals, searches of Google Scholar and Trove identified author accepted manuscripts located in Australian institutional repositories and PubMed Central. A publisher’s notice or statement, either embedded or attached to an author accepted manuscript in an institutional repository, verified the article version and the licensing requirement of the publishers. Data collection also involved the addition of open access data from Web of Science that became available from December 2017. The Web of Science search was re-run to collect data categorising open access articles as Directory of Open Access Journal gold, other gold, other gold-bronze, green accepted, and green published.

The deposit of publication metadata in Australian institutional repositories is an essential part of the Council’s Open Access Policy mandate with data collection involving searches of Google Scholar, Trove, and Australian institutional repositories. The aim was to locate metadata for each article in at least one Australian institutional repository, but where multiple entries existed, data on the entry that included the acknowledgement of the Council and grant numbers had priority. The Australasian Open Access Strategy Group was the source for the list of Australian institutional repositories.

The journal-level information from the article spreadsheet provided the basis for the journal dataset comprising 1,137 titles. A critical function of the journal dataset was the categorisation of titles by journal open access model. While the Directory of Open Access Journals identified full open access journals, the verification of other models involved diverse sources of evidence, including lists of journals and open access statements on journal websites (see Table 2). Resources for locating data on article processing charges included the Directory of Open Access Journals and publishers’ websites. For some hybrid subscription journals, article processing charges varied according to Creative Commons licensing. In the case of journals supported by membership subscriptions that offered hybrid open access, there were different rates for members and non-members. The SHERPA/RoMEO database provided information on each journal’s policy on the deposit of the author accepted manuscript in an institutional repository, as well as the required embargo period and Creative Commons licensing. The journal dataset also recorded the total number of articles and the extent of open access in each title.

Data analysis

The data analysis of the article and journal datasets involved descriptive statistics ranging from simple counts of articles and journal titles to the calculation of proportions expressed as percentages. The built-in statistical and database functions within Microsoft Excel enabled the quantitative analysis of both datasets and the creation of graphical and tabulated forms of results.

The central purpose of the analysis of the article dataset was to determine compliance with the Council’s Open Access Policy. The analysis included counts of open access articles in journals and within the various journal open access models, with proportions expressed as percentages or fractions of the total articles (3,190). The open access data downloaded from the Web of Science database provided a comparison with the data manually collected by this study. The analysis of author accepted manuscripts in Australian institutional repositories included only those articles not found as open access in journals. A further examination determined the extent of author manuscripts in PubMed Central that were neither open access in journals nor in Australian institutional repositories. The scope of metadata in Australian institutional repositories was subject to investigation, especially acknowledgements of the Council and the grant identification number(s).

The data analysis of the journal dataset involved counts of open access articles in each journal and the proportion of open access within the respective journal models, the data then expressed as percentages and fractions of overall journal titles (1,137). The SHERPA/RoMEO data revealed the degree to which journal publishers permitted the deposit of author accepted manuscripts in repositories. Data showing the range of article processing charges enabled the comparison of a select group of publishers. The extent of open access within the Web of Science subject categories was also the focus of analysis.

Results

Funded articles published in 2013 and 2014 achieved compliance of 67.3% with the Council’s Open Access Policy. Open access articles published in journals made up the largest component, comprising 56.24%, with open access in repositories providing a further 11.06%. The pie-chart in Figure 2 provides an outline of overall compliance with the Council’s Policy. The findings were higher than the results in Web of Science that identified 50.69% as open access in journals compared with this study’s 56.24%, and 0.6% as green accepted whereas this study found 11.06%.

Presented in Table 5 are the results of data analysis relating to the number and percentages of open access articles and journal titles in the sample. In column 5, the data on the extent of open access within the respective journal models revealed useful patterns of open access publishing in journals by Council grant recipients. Authors who published in a relatively small number of fully open access journal titles (141), including those published by PLOS and BioMed Central, were responsible for one-fifth (20.85%) of total open access articles. Articles published in PLOS ONE by Council grant recipients contributed 6.6% to open access. While 25.58% of articles were hybrid open access, this represented less than half of the articles published in hybrid subscription journals (67.96%). Authors who published in delayed open access journals contributed 8.75% to total compliance.

The large quantity and complexity of the data on article processing charges prompted the decision to focus on the five publishers responsible for publishing over 50% of the sample, with the results presented in Table 6. The comparatively lower article processing charge of US$1,495 to publish in PLOS ONE was a likely factor contributing to the high level of publishing in that journal by Council grant recipients. While the maximum charges for hybrid open access were high, the minimum rates for a selection of hybrid subscription journals, such as those published by Oxford, were comparable with fully open access journals. An important finding was that many society journals have a range of article processing charges for hybrid open access with as many as six options for members, non-members, and the type of Creative Commons licence. Notwithstanding, over 40% of funded Council articles were not openly accessible on publishers’ websites, with lower levels of non-open access in the journals of some larger subscription publishers. For example, Elsevier (excluding Cell Press) published or co-published 19.12% of the sample but added only 3.26% to overall open access.

The 1,137 journals in the sample encompassed 99 Web of Science subject categories, with hybrid subscription journals represented in all subjects but one. Full open access journals were foremost in the subjects of general and internal medicine, health care sciences, infectious diseases, and multidisciplinary sciences. Many delayed open access journals were evident in critical medicine disciplines. In 43 Web of Science categories, hybrid open access was the only open access publishing option.

While some Council grant recipients deposited author accepted manuscripts in repositories, including 7.24% in Australian institutional repositories and 3.82% in PubMed Central, these results were well below the potential for deposit. The data collected from the SHERPA/RoMEO database indicated that the publishers of most subscription journals allowed the deposit of author accepted manuscripts in institutional repositories. However, this study identified publishers’ conditions that potentially hindered deposit. Elsevier’s barriers on the deposit of author accepted manuscripts in repositories include embargoes of up to 48 months, as well as Creative Commons Attribution-Non Commercial-No Derivatives licensing on all pre-publication versions (Elsevier, 2015).

The study identified metadata in Australian institutional repositories for almost three-quarters (74.92%) of the funded articles. A quarter of the metadata entries included funding acknowledgement of the Council but rarely incorporated the grant identification number. The inclusion of the funding identification number(s) in the metadata is the responsibility of the grant’s chief investigator.

Discussion

The discussion begins with a comparison of funder studies from Australia, Canada, Europe, Spain, the United Kingdom and the United States, the findings of the various investigations outlined in Table 7. Data supplied by Larivière and Sugimoto (2018) for the years 2009–2016 demonstrated that the policies of five funders in the United States and the United Kingdom achieved levels of open access compliance of over 70% and as high as 91.4%. In this study of the National Health and Medical Research Council’s Open Access Policy, over two-thirds (67.3%) of articles published by grant recipients were open access. Open access compliance under the policy of the Spanish Government (Borrego, 2016) was 45.6%, and 38% under the mandate of the Multiple Sclerosis Society of Canada (Bakker, et al., 2017).

According to Larivière and Sugimoto, funders ‘shape compliance through their mandates’ (Larivière and Sugimoto, 2018, p. 486). While access to publicly-funded research is the core principle of most funder mandates, funders differ in their preferred open access models. In the case of funded journal articles, some funders prioritise immediate open access in journals listed in the Directory of Open Access Journals, while others accept open access in hybrid subscription and delayed open access journals. The Wellcome Trust endorses full open access journals as well as journals with agreements with the National Library of Medicine to deposit open access versions in PubMed Central or Europe PubMed Central. Other funders specify the deposit of accepted manuscripts in repositories, especially where the articles are non-open access in journals.

Owing to the different requirements of funder mandates, statistics on the extent of open access in journals vary considerably. Open access in journals under the Council’s Policy amounted to 56.24% compared to 33.9% under the policy of the Multiple Sclerosis Society of Canada (Bakker, et al., 2017). Under the Spanish Government policy, open access in journals was 23.8% (Borrego, 2016) with 18.5% for the Canadian Institutes of Health (Larivière and Sugimoto, 2018). Under funder mandates in the United Kingdom, Larivière and Sugimoto (2018) identified levels of open access in journals ranging from 6.5% for the Wellcome Trust, through to 11.63% for the Medical Research Council, and 12.5% for the Biotechnology and Biological Sciences Research Council.

Data on the journal open access models in which grant recipients publish are useful in describing not only compliance but also patterns of publishing by authors. Borrego (2016) and this study identified the notable contribution of articles published in fully open access journals in meeting funder mandates under the policies of the Spanish Government and the National Health and Medical Research Council. Authors publish in fully open access journals for a variety of reasons including authors’ preference for the model, the high quality of peer review, and the lower article processing charges (Rowley, et al., 2017). A noteworthy parallel between the Spanish and Australian studies was the identification of PLOS ONE as the journal in which most grant recipients published. Observed during this research was the conversion of some journals, such as Nature Communications, to the fully open access model. Notwithstanding, for grant recipients who published in journals covered by over 40 Web of Science medical-related subject categories, there were no fully open access journals, with hybrid the only open access publishing option.

Borrego (2016) and Bakker, et al. (2017) identified low levels of hybrid open access. The reasons for low hybrid open access include the high cost of article processing charges, authors’ uncertainty about the option, and wariness concerning the quality of specific journals (Bakker, et al., 2017; Beall, 2017; Kocher and Kelly, 2016; Poltronieri, et al., 2013; Rowley, et al., 2017; Solomon and Björk, 2012; Zhu, 2017). Jubb, et al. (2017) calculated that more than half of the immediate open access in the United Kingdom in 2016 was hybrid open access (15.4%), a figure much higher than their previous study (Jubb, et al., 2015). This study also found a relatively high level of hybrid open access (25.58%), the likely reason being the use of Council grants for article processing charges under approved projects.

Articles in delayed open access journals in this study, with embargoes expiring within twelve months of publication, contributed 8.75% to compliance with the Council’s Policy. Bakker, et al. (2017) and Borrego (2016) also identified the contribution of delayed open access to overall compliance under funder policies. While there is criticism of embargoes on open access (cOAlition S., 2019c), Haug (2019) argued that many learned societies provide timely and free access to research of vital importance to global health. Jubb, et al. (2017) recognised the longer-term impact of delayed open access at 12 and 24 months after publication. This study found that articles from hybrid and delayed open access journals comprised over one-third (34.33%) of the total open access under the Council’s Policy.

Funder policies implemented by the cOAlition S and the Wellcome Trust do not support delayed open access or the use of grants for the payment of article processing charges for hybrid open access. The criticisms of delayed open access include the lack of immediate open access and the absence of reuse licensing (cOAlition S., 2019c; Wellcome Trust, 2020b). The main censure of hybrid subscription journals is their lack of success in delivering full and immediate open access at reasonable costs (cOAlition S., 2019c). This investigation identified the minimum article processing charges of some hybrid subscription journals as comparable with those charged by fully open access journals (see Table 6). Jubb, et al. (2017) also observed the narrowing of the gap between the article processing charges for hybrid open access and fully open access journals. Most delayed open access journals in this study have hybrid open access options with reduced charges for members.

While a finding of this study was that open access articles in journals comprised 56.24% of total open access, over 40% of Council-funded articles, mostly in hybrid subscription journals, were not openly accessible on publishers’ websites. Bakker, et al. (2017) identified the cost of article processing charges as a deterrent to publishing open access. However, for many researchers, open access is not the primary consideration in selecting journals. Other factors influencing authors’ decisions include the reputation and quality of peer review, as well as individual preferences such as the desire to publish in journals with a high-ranking Impact Factor, although these considerations are not mutually exclusive (Pontika, 2011, 2015; Thornley, et al., 2015; Ware and Mabe, 2015; Zhu, 2017).

Authors who do not publish their articles open access in hybrid subscription journals have the option of depositing author accepted manuscripts in institutional and subject repositories. The results of this study showed that grant recipients contributed 11.06% to the overall extent of open access under the Council’s Policy by depositing in repositories. Notwithstanding, there remained a large quantity of non-open access articles (32.7%) that were non-compliant with the Council’s mandate. To increase open access compliance, the United Kingdom Research Excellence Framework (2019a, 2019b) 2021 policy mandated the deposit of all author accepted manuscripts in institutional repositories, the consequences of which created greater workloads for institutional repository staff and higher stress between researchers and their institutions (Research England, et al., 2018; Ten Holter, 2020). Ten Holter (2020) called for better approaches to meeting open access compliance through institutional repositories. Improved funding and support for institutional repositories are necessary to expand levels of open access (Barbour, 2018; Ten Holter, 2020).

The demand for immediate open access under Plan S and the Wellcome Trust open access policy also has implications for institutional repositories, already struggling with delays on deposit caused by publishers’ embargoes. For example, one-fifth of the non-open access articles in this study were in journals published by Elsevier, some journal-titles with embargoes of up to 48 months, retrospective to 2013 and 2014, and coinciding with the first two years of the Council’s Open Access Policy. Extended and retrospective embargo periods longer than 12 months on depositing in repositories are beyond the time frames set by most funder mandates (Confederation of Open Access Repositories, 2015; Khoo and Lay, 2018). While most funders, including the Council, recommend Creative Commons Attribution licensing, Elsevier, the publisher of most journals in this study, requires Creative Commons Attribution-Non Commercial-No Derivatives licences for all pre-publication versions deposited in institutional repositories (Elsevier, 2015). Funders’ agreements with publishers need to address the issues of embargoes and licensing on the deposit of author accepted manuscripts in repositories.

Repositories play a vital part in the accessibility and visibility of institutional research and fulfilling the mandates of Australia’s leading funders. Notwithstanding, institutional support for repositories in Australia is low and reflected in inadequate funding for repository systems and staffing (Huggard, et al., 2017; Barbour, 2018). Despite the importance of affiliated medical research centres to Australian research, some Australian universities do not support the deposit of publications by centres’ researchers in their institutional repositories (Henty, 2014). While the implementation of Research Information Management Systems at many institutions facilitates the transfer of publication records to repositories (Huggard, et al., 2017), repository staff still need to verify that open access versions in institutional repositories are compliant with publishers’ embargo, copyright and licensing requirements. While authors have the responsibility for depositing the correct open access versions, many are ‘happy for repository staff to deposit on their behalf, but they do not want to be involved in the process’ (Henty, 2014, p. 31). Henty (2014) also uncovered the lack of understanding among authors about the required open access version for deposit in repositories.

This study found a high level (almost 75%) of compliance with the deposit of metadata of funded publications in institutional repositories that is a central component of the Council’s Open Access Policy. Much of the credit for adding metadata is due to repository and library staff manually entering or electronically harvesting from subject repositories and databases (Callan, et al., 2014; Henty, 2014; Organ and O’Hea, 2013). Notwithstanding, less than a quarter of the article metadata acknowledged the Council as the funder, with the grant identification number(s) rarely included. A possible reason for the low funding acknowledgement may be due to repository systems lacking the capability to tag funder and grant details during the period of data collection for this investigation (Huggard, et al., 2017). A more likely explanation is that authors failed to supply funder and grant information, several studies confirming this finding (Álvarez-Bornstein, et al., 2017; Borrego, 2016; Grassano, et al., 2017; Wang, et al., 2015; Wang, et al., 2012). The inclusion of funder and grant identification numbers is a condition of the Council’s funding agreements, with the chief investigator of the grant under the Open Access Policy responsible for ensuring funder acknowledgement information in the publication metadata deposited in institutional repositories.

Institutions need to emphasise the importance of open access to institutional research visibility by addressing the issue of low deposits in institutional repositories. Researchers’ unawareness of author accepted manuscripts and licensing conditions, as well as unfamiliarity with procedures and processes, are among the reasons for levels of depositing that are below potential (Henty, 2014). Institutions need to actively support and promote information programs for researchers that focus on improving open access awareness and knowledge of acceptable open access versions and publishers’ conditions. Required is greater recognition of the work and expertise of institutional repositories in the promotion of institutions’ research and compliance with funder mandates and research evaluation exercises. Adequate funding for infrastructure and resourcing are necessary to ensure that institutional repositories meet international standards for metadata and interoperability, as well as uniform procedures for deposit. For its part, the Council must emphasise among authors, and particularly the chief investigator of grants, that funder acknowledgement and grant identification number(s) in publication metadata are conditions of funding.

A proposed alternative to Australian institutional repositories in meeting compliance with the Council’s Policy is the establishment of an Australasia PubMed Central, but for non-medical funder mandates the focus on biomedicine would be a limitation (Kruesi, et al., 2019a; Kruesi, et al., 2019b). A funder publishing platform, such as Wellcome Open Research, is another option, although this may conflict with non-funder initiatives (Ross-Hellauer, et al., 2018). The National Health and Medical Research Council recognises the network of Australian institutional repositories and Trove for locating funded research outputs (National Health and Medical Research Council, 2015; 2019). Trove provides access to aggregated metadata from multiple institutional repositories through a single platform (Ayres, 2015; Sherratt, 2013). The application of metadata standards across institutional repositories will improve interoperability across systems but requires adequate funding. The promotion of open access also involves greater coordination among funder, institutional, and representative organisations and appropriate support for institutional repositories. Funding bodies need to improve open access awareness for grant recipients through better quality guidelines, procedures, and programmes.

A limitation of this study is the possible exclusion of some funded articles in journals not indexed by Web of Science. Another problem encountered by this research was the removal or relocation of formerly accessible documents and webpages from websites, including funder policies and institutional reports, highlighting the need for systematic approaches to web archiving. A further limitation of this study was the scheduling of data collection at 24 months instead of 12 months after publication required by the Council’s Policy. The extended period caused by time constraints had the potential to inflate the data on author accepted manuscripts in Australian institutional repositories, although the low rate of deposit suggests otherwise. Time also prevented a detailed survey of the policies and procedures of Australian institutional repositories, including interviews with repository managers, to gain a better understanding of the processes involved in the deposit of author accepted manuscripts.

Recommendations for further research include cross-sectional and longitudinal studies into the Council’s Open Access Policy. Comparisons with compliance under the Australian Research Council policy and international funder policies would be useful in providing information on national and global trends. Questionnaires, surveys, and interviews would help gain information on researchers’ attitudes and knowledge of journal open access models and institutional repositories. Further studies need to focus on the crucial roles of institutional repositories and repository staff in meeting compliance. Unpaywall and the open-access feature in Web of Science are valuable tools to scope new projects and refocus existing research, along with Google Scholar and Trove in locating open access versions in Australian institutional repositories. The terminology in use in open access studies requires review to enable cross-analysis and comparison. Further research into open access also needs to include scholarly books, book chapters and conference papers.

Conclusion

This investigation was the first study of compliance with the National Health and Medical Research Council’s Open Access Policy, Australia’s earliest national open access mandate. A key finding of the study was that over two-thirds (67.3%) of the Council’s funded articles were open access. Open access articles published in journals made up the largest component, comprising 56.24%, with repositories contributing a further 11.06%. The findings provide the basis for new research into open access under the Council’s Policy, as well as comparative studies with funder policies both in Australia and internationally.

The National Health and Medical Research Council recognises the advantages of an established network of institutional repositories, with Trove as the central platform, to provide access to Council-funded publications. The deposit of author accepted manuscripts in Australian institutional repositories while low, contributed to open access compliance under the Council’s Policy. Almost one-third of Council-funded articles were not openly accessible and understanding the reasons for non-compliance of this group is critical. Factors likely to contribute to low deposit rates include the lack of author awareness of acceptable open access versions and publishers’ conditions on deposit. Improving institutional repository deposit rates is an important avenue to increase open access and compliance with funder mandates.

Acknowledgements

The authors thank the reviewers for their helpful comments on the original paper. Noreen Kirkman is grateful for the support of the Australian Government’s Research Training Program Scholarship.

About the author

Noreen Kirkman completed a master’s thesis in 2018 on the National Health and Medical Research Council’s Open Access Policy through the discipline of Libraries, Archives, Records and Information Science, Curtin University (Perth, Australia). Email: noreen.kirkman@gmail.com
Gaby Haddow is Associate Professor and leads the discipline area Libraries, Archives, Records and Information Science at Curtin University. Her research focuses on scholarly communication, research evaluation for the Humanities, Arts and Social Sciences, and research support in academic environments. Email: g.haddow@curtin.edu.au