A qualitative interview study to determine barriers and facilitators of implementing automated decision support tools for genomic data access

A qualitative interview study to determine barriers and facilitators of implementing automated decision support tools for genomic data access | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article A qualitative interview study to determine barriers and facilitators of implementing automated decision support tools for genomic data access Vasiliki Rahimzadeh, Jinyoung Baek, Jonathan Lawson, Edward S. Dove This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-3849259/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 05 May, 2024 Read the published version in BMC Medical Ethics → Version 1 posted 8 You are reading this latest preprint version Abstract Data access committees (DAC) gatekeep access to secured genomic and related health datasets yet are challenged to keep pace with the rising volume and complexity of data generation. Automated decision support (ADS) systems have been shown to support consistency, compliance, and coordination of genomic data sharing of data access review decisions. However we lack understanding of how DAC members perceive the value add of ADS, if any, on the quality and effectiveness of their reviews. In this qualitative study, we report findings from 13 semi-structured interviews with DAC members from around the world to identify relevant barriers and facilitators to implementing ADS for genomic data access management. Participants generally supported pilot studies that test ADS performance for example in cataloging data types, verifying user credentials and tagging datasets for use terms. Concerns related to over-automation, lack of human oversight, low prioritization, and misalignment with institutional missions tempered enthusiasm for ADS among the DAC members we engaged. Tensions for change in institutional settings within which DACs operated was a powerful motivator for why DAC members considered the implementation of ADS into their access workflows, as well as perceptions of the relative advantage of ADS over the status quo. Future research is needed to build the evidence base around the comparative effectiveness and decisional outcomes of institutions that do/not use ADS into their workflows. Figures Figure 1 1. Introduction Genomics is among the most data-prolific scientific fields and is expected to surpass the storage needs and analytic capacities of Twitter, YouTube, and astronomy combined by as soon as 2025 1 . To meet rising demands for genomic data and their efficient collection and use, national genomics initiatives 2 rely on largescale repositories to pool data resources and incentivize data sharing 3–5 . The ‘data commons’ model has since become the flagship approach for many of these initiatives 6 , and prioritizes research collaboration and data access over proprietary exclusion in the data 3 . Data access committees (DACs) are principally charged with ensuring only bona fide researchers conducting research permitted by participants’ informed consent are approved to access the data 7 . As such, DACs form the backbone of responsible data sharing ecosystems since they govern access to and compliant use of genomic and other health data 8–10 . However, DACs may not contribute to efficient data access provisions as effectively as other review models may allow 11 . In the standard model of data access review, DACs manually review a data requester’s application and assess it against pre-defined criteria. Criteria may include appropriateness of the data requested, data use terms set by data providers, and data privacy and security requirements set by the institution and by law 7 . As with most, if not all, human-mediated activities, manual review of these criteria can be a laborious and error-prone process. For example, DACs may interpret language describing permitted data uses differently, and the terms themselves can sometimes be ambiguous 12 . Faced with this ambiguity, DACs are forced to make subjective judgments about whether requests for data access truly align with permitted data uses, if they have been preserved following many data exchanges or exist at all. Inconsistencies in how data use terms are articulated in consent forms and subsequently interpreted and executed by DACs across the biomedical ecosystem 12 can lead to delayed and inconsistent data access decisions, and risk violating the terms by which patients or participants contributed their data in the first place. Other steps in the data access pipeline can also contribute to research delays. Emerging research suggests there is growing inefficiency, inconsistency, and error in the manual, entirely human-mediated review of data access agreements 11,13 which are executed in finalizing approved data access requests. Many researchers furthermore still rely on the traditional method of copying-and-downloading data once approved. The copy-download approach multiplies security risks 9 , and is quickly becoming unreasonable given the expanding size and complexities of genomic datasets 14,15 . Standards’ developers and software engineers have therefore sought to semi-automate three axes of data access control within cloud environments – user authentication, review of access requests, and concordance of the proposed research with the data use terms of the data requested 12 . Automated decision support (ADS) systems are a coordinated system of algorithms, software, and ontologies 16 that aid in categorizing, archiving, and/or acting on decision tasks for data access review. The Data Use Oversight System (DUOS) typifies one such automated decision support 17 . In recent beta tests, DUOS was successfully shown to concur 100% of the time with human-decided access requests 13 , and also codifies 93% of genomic datasets in NIH’s dbGaP 18 . While ADS can supplement human-mediated DACs with semi-automated technical solutions, no systematic investigation has sought to characterize relevant barriers and facilitators to ADS in practice 19 . Moreover, we lack understanding of how DAC members perceive the value added by ADS, if any, on the quality and effectiveness of data access review decisions, as well as what challenges they anticipate in adopting ADS taking into account the myriad organizational structures within which DACs operate. Now is an opportune time to study the implementation barriers and facilitators to using ADS solutions for data access as their development is converging with large-scale data migration to the cloud and hold the promise of near-instant data access decisions. The genomics community can learn important lessons from previous attempts at (premature) ADS implementation without purposeful stakeholder engagement in public health 20 , law enforcement 21 and in clinical care 22 . In this article, we report empirical findings on the “constellation of processes” relevant for implementing ADS for genomic data access management and provide practical recommendations for institutional data stewards that are considering or have already implemented ADS in this context. 2. Methods We conducted a qualitative description study that engaged prospective end users of ADS for genomic data governance to explore What are the barriers and opportunities of implementing automated workflows to manage access requests to genomic data collections, and what effect do ADS have on DAC review quality and effectiveness? We adopted Damshroder and colleagues’ definition of implementation as the “critical gateway between an organizational decision to adopt an intervention and the routine use of that intervention” 23 in order “study the constellation of processes intended to get an intervention into use within an organization” 23 . We applied the Consolidated Framework for Implementation Research (CFIR) to compare genomic data access processes and procedures across to better understand implementation processes for automated workflows to manage genomic data access across international, publicly funded genomic data repositories. The CFIR provides a “menu of constructs” associated with five domains of effective implantation which have been rigorously meta-theorized—that is, synthesized from many implementation theories (Fig. 1 ). In addition, the CFIR provides a practical guide to systematically assess potential barriers and facilitators ahead of an innovation’s implementation (L. Damschroder et al. 2015). The CFIR is also easily customizable to unveiling bioethical issues during implementation in genomics and has been applied in prior work (Burke and Korngiebel, 2015; Smit et al., 2020). The interview guide was developed specifically for this study and is available in Supplementary Materials 2. <> 3. Data collection We conducted a total of 13 semi-structured interviews with 17 DAC members between 27 April and 24 August 2022. Prospective interviewees indicated their interest in being invited to a follow up interview following their participation in survey published elsewhere 24 . All interviews were conducted virtually and audio/video recorded on Zoom. We used validated interview guides from the official CFIR instrument repository ( https://cfirguide.org/evaluation-design/qualitative-data/ ) to probe the barriers and opportunities of implementing ADS solutions for DAC review of data access requests. Interviewees were also recruited from the Data Access Committee Review Standards Working Group (DACReS WG) chaired by authors VR, JL, and ESD, as well as from an internet search of publicly funded genomic data repositories worldwide. 4. Data analysis We first applied a deductive coding frame to the interview transcripts based on a framework analysis approach (Pope, Ziebland, and Mays 2000) and the publicly accessible CFIR codebook available in the Supplemental Materials 1 . To ensure the reliability of conclusions drawn, two independent reviewers (VR and JB) tested the coding schema on three transcripts until reaching a recommended interrater reliability score of 0.83 before analyzing the remaining qualitative dataset. All coding discrepancies during the coding pilot were resolved by consensus discussion. <> Table 1 CFIR Code Application Results High frequency >=25 Medium frequency 10–25 Low frequency < 10 Code (number of code applications) Tension for Change (98), Relative Advantage (72), Knowledge &Beliefs about the Innovation (47), Structural Characteristics (36), Planning (33), Cosmopolitanism (32), External Policy & Incentives (30), DAC tools ( code created by the team) (30), Compatibility (30), Needs & Resources of those Served by the Organization (27), Key Stakeholders (25) Culture (23), Networks & Communications (22), Relative Priority (21), Cost (21), Adaptability (21), Innovation Source (20), Reflecting & Evaluating (19), External Change Agents (18), Formally Appointed Internal Implementation Leaders (17), Available Resources (16), Individual Identification with Organization (15), Access to Knowledge & Information (14), Evidence Strength & Quality (14), Peer Pressure (12), Other Personal Attributes (12), Opinion Leaders (11) Individual Stage of Change (9), Goals & Feedback (9), Champions (9), Implementation Climate (8), Engaging (8), Self-efficacy (7), Leadership Engagement (7), Complexity (6), Trialability (6), Learning Climate (6), Characteristics of Individuals (3), Innovation Participants (3), Readiness for Implementation (2), Executing (2), Design Quality & Packaging (1), Process (1) 5. Results Geographical, Institutional, and Demographic Background of Participants Forty-one percent of interviewees worked within U.S.-based DACs, while the remaining 59% of interviewees represented DACs at institutions in Canada, the U.K., Spain, Tunisia, Australia, and Japan (Table 1 ). Nearly 60% of interviewees worked at a non-profit research institute, 24% represented an academic-affiliated research institution, 12% represented a government research agency, and 6% were affiliated with a research consortium. Seventy-six percent of interview participants identified as female, and 24% as male. a. Opportunities for ADS We categorized the frequency of CFIR implementation factors referenced in our interviews in Table 2 . Our findings suggest that there are three major facilitators to implementing ADS for genomic data governance: 1) external policy and need for efficient workflows, 2) institutional ability to scale the ADS, and 3) interoperability. Table 2 Participant demographics DAC location Total % (n) United States 54 (7) Canada 12 (2) United Kingdom 6 (1) Spain 6 (1) Tunisia 6 (1) Australia 18 (3) Japan 12 (2) Institutional type Non-profit Research Institute 59 (10) 1) External policy and need for efficient workflows Participants considered adopting ADS to comply with new data sharing mandates from research funders (e.g. National Institutes of Health) and those imposed by peer reviewed journals. The demand for and scope of compliant data access review has had a ripple effect on ethics oversight bodies 25 , including DACs, as a result of these new requirements 24 . Most DAC members we engaged currently perform their reviews manually. Members review all data access requests individually or as a committee and make decisions on each request received in the order they were received. Given the anticipated increase in the number of data access requests 26 , our participants noted the reduced workload and costs associated with ADS could contribute to better review efficiencies, without a concomitant loss in review quality and risk of noncompliance with data use conditions. We found that participants perceived ADS could reduce DAC member workload by streamlining the intake process for data access requests and verifying that the request matched the terms of use in the original consent obtained at data collection. Indeed, participants noted the initial screening of Data Access Requests (DARs) was a common rate-limiting step. DACs often begin the review process by verifying that all necessary information is documented in the request (e.g. study purpose, datasets requested, ethics review). This step can be time-consuming because the requirements can vary depending on the researcher’s institution and the datasets they request. We requested that participants share a copy of their DAR form before, during, or after the interview to compare what information DACs typically required to process a DAR. We found the form fields as well as length of the DAR (from 3 to 18 pages) differed considerably from DAC to DAC. Our participants believed that this is where ADS could be useful by automatically flagging missing information and documents, verifying the authenticity of a requester’s identity and the submitted documents, and then sending notifications to requesters if more information is needed. As one interviewee put it: Because one of the biggest concerns in our DAC is that sometimes it takes too much time to be read by all the nine members. … They’re institutional directors or university professors. So I think it will help. Maybe if you have 50% of the work done by an automated system, so you just have to do the 50%. I think … this will be a good motivation for them saying ‘OK’ [to implement ADS]. Participant M 2) Scalability Participants also believed ADS-enabled workflows could be scalable, cost-effective solutions to management of not just newly generated data, but also for legacy data when grant funding ends because ADS can easily store and quickly present data use conditions and audit past DAC reviews: Actually there are lots of costs related to data sharing, particularly if I'm sharing data from the 1990s, for example. I don't have any money or budget anymore to prepare the data [for secondary uses]. … And similarly, when it comes to these reports [on data sharing activities], there's no extra money for doing the work to create those reports. But we're having to report back over assets from years, decades in fact. And there was always just a little bit of a hint ‘oh well, maybe we'll find some money’. No, no, you have to find it out on your own. Participant F Some participants also shared that ADS could be helpful when DAC members or data generators leave the institution, disrupting review continuity consistency. Unlike for large, well-funded government repositories, many DACs at smaller institutions lack resources for long-term data preservation and access management for data of increasing complexity and volume: As the program scales, the participant diversity scales, the data diversity scales. I think it is almost impossible to see a scenario where we do not rely on some level of automation to support human decision making about what is responsible use. Participant J I mean potentially as we grow over the years, you know what's going to happen. … we've also discussed some scenarios, where, for example, we find ourselves with a larger amount of requests coming in, [and] we only accept applications up to certain days and then, we open this next quarter, close it again. But there potentially could be room for automation depending on the increase in request in the coming years. Participant A 3) Interoperability According to DAC members we interviews, ADS tools could provide centralized, interoperable solutions to facilitate inter-organizational and international data sharing. Participants perceived that ADS could motivate use of standardized request forms, access agreements, dataset identifiers, and methods for verifying researcher identities: “But this [ADS] will free up a lot of time in the process is it also potentially means that it will become easier for, if you're working in a team to hand off tasks as well because you will have a single system. … Also, consistency between organizations. If we have multiple organizations take this up, it's going to mean less lead time. [Let’s] say people take a new job in a new place. We'll actually have some software that people will recognize and be able to use and uptake, which we've been trying to go towards without ethics approval processes within the hospital and health services… [standardized] systems makes it easier for actual communication between organizations on processes, because everyone kind of begins to know what's happening. Participant E b. Barriers to implementing ADS Despite clear advantages of ADS for genomic data access management, our interviewees identified significant barriers to implementation within DAC workflows, including: 1) lower priority compared to more immediate governance challenges, 2) ill equipped personnel and structures within the institution, 3) costs, and 4) degree of human oversight. 1) Low priority Many participants reported that institutional leadership prioritized other competing research data needs over investing in new data governance structures (e.g. generating quality data, increasing diversity in datasets, collaborating with underrepresented groups of researchers and participants, and releasing datasets). Participants believed researchers in general understand why quality and effective review of data access is important for responsible genomic data sharing but are firstly concerned with data quality. Another suspected reason that ADS implementation ranked lower on institutional priorities was that there had not yet been a significant data incident. As one participant put it: I don’t think that the program thinks it is a very high priority to streamline any of the [data access oversight] process. I think that it will either take something bad happening and then realizing that we need additional capacities on [DAC], or some other hiccup to really promote that need. Participant O Because budgets for data governance are not always included in grants, researchers may be less motivated to invest in the additional, largely unpaid work related to data governance. Insufficient resourcing for data sharing and governance mechanisms prospectively in research study design inevitably challenge the downstream execution of data governance upon deposit of the research data once generated, according to at least one DAC member we interviewed: We found that some people don’t prioritize [data governance] because it’s not helpful to them, because it’s not our primary function as a department. You know, we’re producing new data. That’s usually what people, researchers are doing. They’re not thinking about what happens to their old data. So, it’s not much of a priority. Having said that, research funders are getting very keen for us to use their data. So, there is that sort of tug [of war]. … If I go into a senior team meeting, you know, something else will be the priority. Participant F 2) Structural characteristics of an organization We also found a close correlation between several structural characteristics of the institution (e.g. years in operation, number of personnel, and database size) and participants’ perceived barriers to ADS implementation. For instance, many participants served on DACs that were established within the last 1–3 years coinciding with the creation of the institution’s database. As the datasets grow, and more researchers are attracted to the resource, there is greater potential for existing management processes to be overwhelmed. It is precisely at this early juncture that DACs weight their options for ADS tools, and proactively address relevant barriers ahead of future plans for implementation. Some DAC members preferred to gain more experience with existing data access management in these early years of data release before integrating ADS “because we’re not sure how [name of participant’s country] citizens feel or consider about the automatic decision on data sharing.” Participant K. 3) Cost While cost was not a primary concern for ADS implementation at well-funded big data repositories, it was a significant barrier for DAC members working at smaller repositories, individual research departments, or research programs associated with a genomics consortium who were more often supported by research grants or contracts rather than an independent funding source. “We [data governance office] are supported through project-specific funding. … Governance ends up being a little bit of this indirectly supported component of our work and services. That has limited the ways in which we can innovate around governance. … We don’t have a huge budget.” Participant N Without dedicated budget for human and material resources, some DAC members were concerned that the initial investment in ADS and significant changes to current workflows would be key issues, to say nothing of new education and training materials and updates to internal policies, among other ancillary revisions to internal workflows. 4) Lack of human oversight While some DAC members were enthusiastic about improvements in efficiency and consistency of ADS, participants unanimously rejected the idea of fully automating access management: “no matter what we do with automation that I feel there always needs to be that human element who's coming in and checking. So, there will always be that barrier to upscaling” Participant E. Other participants emphasized that prior to implementation, they would need to gauge how research participants at their own institution as well as the general public would react to ADS for data access review. Participants were also skeptical that ADS could adequately assess complex, sensitive data reuse issues which they felt required a deep understanding of ethical, legal, and sociocultural contexts within which data were collected, used, and shared. Some DAC members reported asking data requesters to clarify their study purpose and justify their need for specific datasets in recognition of these sociocultural dimensions. I’m also someone who thinks that it's important to be very critical about what's the nature of the work being done. Maybe it's solid from a scientific point of view. But are there other concerns from other perspectives that need to be taken into account? That is partly why we have community members on the [committee], and that's something I'm not sure can be simplified or automated.” Participant B However, when it comes to automating anything that requires reviewing information where there might be a lot of nuances, where there might be a lot of interpretation that's required, I'm a little bit more hesitant simply because I think to some extent you do need some room for a little bit of mulling over the information, … and I think there are some information that come through with requests, that don't neatly fit into check boxes. Participant B 6. Discussion Participants overall perceived that ADS tools could be well positioned to help DACs streamline data access compliance. While believed to beneficial, ADS solutions were unlikely to immediately or directly advance the research organization’s core mission (e.g. collecting quality data and driving scientific discoveries and innovations) according to the DAC members we engaged. Participants also tended to regard ADS implementation, as well of data governance workflow solutions as a comparatively low priority compared to regulatory compliance, investigator support, and database curation, among other competing demands on DAC member time. Most research grants allow investigators to apply for support for data collection and analysis, but rarely to establish actual governance structures needed to stand up access management services. DAC members therefore shared that it can be difficult to convince institutional leadership to invest in data governance solutions such as ADS. Some participants noted that the real benefit of ADS is to streamline workflows that are currently time consuming and tedious, for example preliminary review of applications to ensure there are no missing elements. Applying ADS to facilitate these workflows could free DAC members to dedicate more time to deliberate on more substantive ethics issues raised by data access requests. While data governance has often been considered auxiliary work, new research findings and new U.S. federal government policies, such as the National Institutes of Health Data Management and Sharing policy (DMS), have elevated its importance by placing additional requirements for data sharing 27 . Some of our study participants mentioned that they believe researchers are drawn to their databases not because of their data access policies and practices, but because of the quality and diversity of their datasets. This perspective differs from what another study found suggesting that ease of access is at least marginally important when choosing databases/institutions for research data 28 . The DMS policy will accelerate the accumulation of an enormous number of datasets, which will, absent interventions, significantly push up the costs associated with data storage and management. If not developed in collaboration, institutional databases created for the explicit purpose of sharing research data generated from federal funds has the potential to introduce myriad access pathways that make the data effectively ‘shared’ but unfindable. One of our participants mentioned that some legacy data, which date back to the 50s and 60s, were transferred to their organization due to a closure of the repositories which originally stored these datasets. This case suggests that there is an urgent need for more efficient and sustainable solutions for data access management and sharing and perhaps is reasonable cause to ensure that there is a contingency plan for publicly-funded data shared via non-publicly supported repositories in the event the repository closes or changes in policy or personnel. Limitations Our results should be considered in light of several methodological limitations. While geographically diverse, many of our interview participants were affiliated with DACs based at large, well-resourced research institutions. It is likely that responses and perceptions of implementation factors related to ADS would differ substantially if more DACs from low- or under-resourced institutions were represented in our sample. Our data collection design relies on self-reports of institutional data access policy and procedures. Many interview participants were aware of the Global Alliance for Genomics and Health, and the data access committee review standards we were principally involved in developing 29 . Thus while we endeavored to create a safe, open environment for participants to share their honest views, social desirability bias related to our prior work may have influenced how participants responded. Lastly, CFIR predefines sociological constructs relevant to implementation. Our analysis was therefore limited only those constructs covered in the framework when others might have emerged inductively if we adopted an alternative analytic frame. Conclusion In this article, we reported findings from semi-structured qualitative interviews with DAC members from around the world on the relevant barriers and facilitators of implementing ADS for genomic data access management. Our findings suggest there is general support for pilot studies that test ADS performance in certain elements of the data access management workflows, such as cataloging data types, verifying user credentials and tagging datasets for use terms. Participants indicated that ADS should supplement, but not replace DAC member work. This sentiment was especially strong with respect to tasks that were perceived to require sensitivity and human value-judgments such as privacy protections, group harms, and study purpose. Rather, our findings offer cautious optimism regarding the ways in which algorithms, software, and other machine-readable ontologies can streamline aspects of DAC decision-making while also enabling new opportunities for improving consistency and fairness in DAC decisions. To that end, we conclude with practical recommendations for institutional data stewards that are considering or have already implemented ADS for data access management. The explosion in the volume and complexity of genomic and associated health data is converging with the need to manage access more efficiently to these data. Such trends point intuitively to solutions that can help to alleviate, or at least prevent bottlenecks in the access process to preserve the scientific and social value of data generated from public investments in research. To put ADS solutions to the test, future research should compare access decisions and their outcomes between institutions who do/not use such tools for data access management; whether ADS delivers on its efficiency promises and liberates DAC member time previously spent addressing procedural matters and instead create more opportunities for committee deliberation on substantive ethics issues. Declarations Acknowledgements The authors wish to thank members of the Data Access Committee Review Standards Working Group, and the Regulatory and Ethics Work Stream of the Global Alliance for Genomics and Health for their contributions to the intellectual community that inspired this work. Author Contributions Authors VR, JL and ED conceptualized, designed, and carried out the study. Author JB led in the data collection and analysis and drafting of early manuscript drafts. All authors, VR, JB, JL and ED took part in writing and editing the manuscript, responding to peer reviewer comments and approved the final version. Ethics approval This study was reviewed and approved the Stanford University Institutional Review Board. All participants were informed of the purpose of the study, funding, risks and benefits at the time of invitation. A research assistant consented all participants prior to the interview, and were provided opportunities to ask any questions. Competing Interests All authors are members of the Global Alliance for Genomics and Health. JL is lead software engineer of the Data Use Ontology and is a member of the Broad Institute Data Access Committee. Consent for publication Not applicable. Availability of data and materials Materials described in the manuscript and data supporting our findings can be made available upon request. All requests should be directed to Vasiliki Rahimzadeh, PhD at [email protected] . References Stephens ZD, Lee SY, Faghri F, Campbell RH, Zhai C, Efron MJ, Iyer R, Schatz MC, Sinha S, Robinson GE. Big Data: Astronomical or Genomical? PLoS Biol. 2015;13:e1002195. 10.1371/journal.pbio.1002195 . IQVIA Institute for Human Data. Science Understanding the Global Landscape of Genomic Initiatives: Progress and Promise. Cook-Deegan R, McGuire AL. Moving beyond Bermuda: Sharing data to build a medical information commons. Genome Res. 2017;27:897–901. 10.1101/gr.216911.116 . Evans B. (2019). Genomic data commons. Governing Privacy in Knowledge Commons, 75–105. Contreras JL, Knoppers BM. The genomic commons. Annu Rev Genom Hum Genet. 2018;19:429–53. 10.1146/annurev-genom-083117-021552 . Grossman RL. Ten lessons for data sharing with a data commons. Sci Data. 2023;10:120. 10.1038/s41597-023-02029-x . Cheah PY, Piasecki J. Data Access Committees. BMC Med Ethics. 2020;21:1–8. 10.1186/s12910-020-0453-z . Shabani M, Thorogood A, Borry P. Eur J Hum Genet. 2016;24:1671–5. 10.1038/ejhg.2016.111 . Who should have access to genomic data and how should they be held accountable? Perspectives of Data Access Committee members and experts. Ramos EM, Din-Lovinescu C, Bookman EB, McNeil LJ, Baker CC, Godynskiy G, Harris EL, Lehner T, McKeon C, Moss J, GAIN Data Access Committee. Am J Hum Genet. 2013;92:479–88. 10.1016/j.ajhg.2012.08.034 . A Mechanism for Controlled Access to GWAS Data: Experience of the. Shabani M, Knoppers BM, Borry P. EMBO Mol Med. 2015;7:507–9. 10.15252/emmm.201405002 . From the principles of genomic data sharing to the practices of data access committees. Mello MM, Triantis G, Stanton R, Blumenkranz E, Studdert DM. Waiting for data: Barriers to executing data use agreements. Science. 2020;367:150–2. 10.1126/science.aaz7028 . Lawson J, Cabili MN, Kerry G, Boughtwood T, Thorogood A, Alper P, Bowers SR, Boyles RR, Brookes AJ, Brush M, et al. The Data Use Ontology to streamline responsible access to human biomedical datasets. Cell Genomics. 2021;1:100028. 10.1016/j.xgen.2021.100028 . Lawson J. Empirical Validation of an Automated Approach to Data Use Oversight. Cell Genomics Forthcoming; 2021. Van der Auwera G, O’Connor B. Genomics in the Cloud: Using Docker, GATK, and WDL in Terra. 1st ed. O’Reilly Media; 2020. Birger C, Hanna M, Salinas E, Neff J, Saksena G, Livitz D, Rosebrock D, Stewart C, Leshchiner I, Baumann A, et al. FireCloud, a scalable cloud-based platform for collaborative genome analysis: Strategies for reducing and controlling costs. (Bioinformatics). 2017. 10.1101/209494 . Rahimzadeh V, Lawson J, Rushton G, Dove ES. Leveraging Algorithms to Improve Decision-Making Workflows for Genomic Data Access and Management. Biopreserv Biobank. 2022. 10.1089/bio.2022.0042 . Data Use Ontology. (2022). Lawson J, Ghanaim EM, Baek J, Lee H, Rehm HL. Aligning NIH’s existing data use restrictions to the GA4GH DUO standard. Cell Genomics. 2023;3:100381. 10.1016/j.xgen.2023.100381 . Rahimzadeh V, Lawson J, Baek J, Dove ES. Automating Justice: An Ethical Responsibility of Computational Bioethics. Am J Bioeth. 2022;22:30–3. 10.1080/15265161.2022.2075051 . Obermeyer Z, Powers B, Vogeli C, Mullainathan S. Dissecting racial bias in an algorithm used to manage the health of populations. Science. 2019;366:447–53. 10.1126/science.aax2342 . Hartzog W, Conti G, Nelson J, Shay LA. Inefficiently Automated Law Enforcement. Mich State Law Rev. 2016;1763:1763–96. Ledford H. Millions affected by racial bias in healthcare algorithm. Nature. 2019;574:608–10. Damschroder LJ, Aron DC, Keith RE, Kirsh SR, Alexander JA, Lowery JC. Fostering implementation of health services research findings into practice: a consolidated framework for advancing implementation science. Implement Sci. 2009;4. 10.1186/1748-5908-4-50 . Lawson J, Rahimzadeh V, Baek J, Dove ES. (2023). Achieving Procedural Parity in Managing Access to Genomic and Related Health Data: A Global Survey of Data Access Committee Members. Biopreservation and Biobanking, bio.2022.0205. 10.1089/bio.2022.0205 . Rahimzadeh V, Serpico K, Gelinas L. Institutional review boards need new skills to review data sharing and management plans. Nat Med. 2023;1–3. 10.1038/s41591-023-02292-w . Cabili MN, Lawson J, Saltzman A, Rushton G, O’Rourke P, Wilbanks J, Rodriguez LL, Nyronen T, Courtot M, Donnelly S, et al. Empirical validation of an automated approach to data use oversight. Cell Genomics. 2021;1:100031. 10.1016/j.xgen.2021.100031 . Final NIH. Policy for Data Management and Sharing https://grants.nih.gov/grants/guide/notice-files/NOT-OD-21-013.html . Trinidad MG, Ryan KA, Krenz CD, Roberts JS, McGuire AL, De Vries R, Zikmund-Fisher BJ, Kardia S, Marsh E, Forman J, et al. Extremely slow and capricious: A qualitative exploration of genetic researcher priorities in selecting shared data resources. Genet Sci. 2023;25:115–24. 10.1016/j.gim.2022.09.003 . Global Alliance for Genomics and Health. (2021). Data Access Committee Guiding Principles and Procedural Standards Policy. Additional Declarations Competing interest reported. All authors are members of the Regulatory and Ethics Work Stream of the Global Alliance for Genomics and Health. JL is lead software engineer of the Data Use Ontology and is a member of the Broad Institute Data Access Committee. Supplementary Files SupplementalMaterials1.docx BMCMedEthicsSupplementarymaterials2.docx Cite Share Download PDF Status: Published Journal Publication published 05 May, 2024 Read the published version in BMC Medical Ethics → Version 1 posted Editorial decision: Revision requested 01 Apr, 2024 Reviews received at journal 29 Mar, 2024 Reviews received at journal 02 Feb, 2024 Reviewers agreed at journal 22 Jan, 2024 Reviewers invited by journal 22 Jan, 2024 Editor assigned by journal 16 Jan, 2024 Submission checks completed at journal 16 Jan, 2024 First submitted to journal 09 Jan, 2024 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-3849259","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":267338722,"identity":"85ce59ee-3611-4be9-804a-25fff46a8e68","order_by":0,"name":"Vasiliki Rahimzadeh","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABFUlEQVRIiWNgGAWjYJCCAw8MEniANBtDApjLw/gYIpGAW0sCmhZmY0JagFJgSTaoCTxs0vi0GBw//vBAQkGaDIP04WcPHubUyfMd7z1WXVCzjYGfPccAq5YzOQZAh+XwMPClmRskbmMznHnmXNrtGcduM0j2vMGqRbIhB+SXCh4GHgYzicRtPIwbbuSY3eZtuM1gcAO7LZL9zx9AtbB/A2qRsN9w/41ZMUiLPQ4t/BIJUIfx8IBsMUjccIPHjBlsiwQuLW9AWtJ42Hh4yoBaEpJnnskxlgb6hUfizLMCbFrY+NMff/jwJ9men4d9m+TPbXW2fcfPGH4uqLktx9+evAFrKMP1ogvw4FU+CkbBKBgFowAvAACn0WDclCyGewAAAABJRU5ErkJggg==","orcid":"","institution":"Baylor College of Medicine","correspondingAuthor":true,"prefix":"","firstName":"Vasiliki","middleName":"","lastName":"Rahimzadeh","suffix":""},{"id":267338723,"identity":"89ecc11a-9f96-4306-9044-35986e5bbce1","order_by":1,"name":"Jinyoung Baek","email":"","orcid":"","institution":"Broad Institute","correspondingAuthor":false,"prefix":"","firstName":"Jinyoung","middleName":"","lastName":"Baek","suffix":""},{"id":267338724,"identity":"7f6e0dc4-6e7a-40e8-895b-11d630ea84d2","order_by":2,"name":"Jonathan Lawson","email":"","orcid":"","institution":"Broad Institute","correspondingAuthor":false,"prefix":"","firstName":"Jonathan","middleName":"","lastName":"Lawson","suffix":""},{"id":267338725,"identity":"218b37e0-55c7-4695-bfde-7a88bca97e3a","order_by":3,"name":"Edward S. Dove","email":"","orcid":"","institution":"University of Edinburgh","correspondingAuthor":false,"prefix":"","firstName":"Edward","middleName":"S.","lastName":"Dove","suffix":""}],"badges":[],"createdAt":"2024-01-09 22:14:11","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-3849259/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-3849259/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1186/s12910-024-01050-y","type":"published","date":"2024-05-05T12:18:03+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":49776886,"identity":"bbb5066d-7e81-46fe-acf6-1bbf3c36d2d4","added_by":"auto","created_at":"2024-01-17 21:14:10","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":218221,"visible":true,"origin":"","legend":"\u003cp\u003eConsolidated Framework for Implementation Research and associated constructs applied to genomic data access management.\u003c/p\u003e","description":"","filename":"F1.png","url":"https://assets-eu.researchsquare.com/files/rs-3849259/v1/466b39d1fa18b5933405d689.png"},{"id":55871096,"identity":"7c39a991-183d-4b30-9bf9-0c9b64b9f4df","added_by":"auto","created_at":"2024-05-05 12:18:11","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":733978,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-3849259/v1/12d845ca-ade0-496b-989f-e7e32f1bf08f.pdf"},{"id":49776888,"identity":"bdd22ea0-7562-4473-aa2d-156a4d34d228","added_by":"auto","created_at":"2024-01-17 21:14:10","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":49014,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementalMaterials1.docx","url":"https://assets-eu.researchsquare.com/files/rs-3849259/v1/4e30504461a5eccb521602e2.docx"},{"id":49776887,"identity":"d72491b4-cb72-4c4f-9118-50589471fe42","added_by":"auto","created_at":"2024-01-17 21:14:10","extension":"docx","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":24325,"visible":true,"origin":"","legend":"","description":"","filename":"BMCMedEthicsSupplementarymaterials2.docx","url":"https://assets-eu.researchsquare.com/files/rs-3849259/v1/5f90b893941d6b53510b5ac1.docx"}],"financialInterests":"Competing interest reported. All authors are members of the Regulatory and Ethics Work Stream of the Global Alliance for Genomics and Health. JL is lead software engineer of the Data Use Ontology and is a member of the Broad Institute Data Access Committee.","formattedTitle":"A qualitative interview study to determine barriers and facilitators of implementing automated decision support tools for genomic data access","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eGenomics is among the most data-prolific scientific fields and is expected to surpass the storage needs and analytic capacities of Twitter, YouTube, and astronomy combined by as soon as 2025\u003csup\u003e1\u003c/sup\u003e. To meet rising demands for genomic data and their efficient collection and use, national genomics initiatives\u003csup\u003e2\u003c/sup\u003e rely on largescale repositories to pool data resources and incentivize data sharing\u003csup\u003e3\u0026ndash;5\u003c/sup\u003e. The \u0026lsquo;data commons\u0026rsquo; model has since become the flagship approach for many of these initiatives\u003csup\u003e6\u003c/sup\u003e, and prioritizes research collaboration and data access over proprietary exclusion in the data\u003csup\u003e3\u003c/sup\u003e. Data access committees (DACs) are principally charged with ensuring only \u003cem\u003ebona fide\u003c/em\u003e researchers conducting research permitted by participants\u0026rsquo; informed consent are approved to access the data\u003csup\u003e7\u003c/sup\u003e. As such, DACs form the backbone of responsible data sharing ecosystems since they govern access to and compliant use of genomic and other health data\u003csup\u003e8\u0026ndash;10\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eHowever, DACs may not contribute to efficient data access provisions as effectively as other review models may allow\u003csup\u003e11\u003c/sup\u003e. In the standard model of data access review, DACs manually review a data requester\u0026rsquo;s application and assess it against pre-defined criteria. Criteria may include appropriateness of the data requested, data use terms set by data providers, and data privacy and security requirements set by the institution and by law\u003csup\u003e7\u003c/sup\u003e. As with most, if not all, human-mediated activities, manual review of these criteria can be a laborious and error-prone process. For example, DACs may interpret language describing permitted data uses differently, and the terms themselves can sometimes be ambiguous\u003csup\u003e12\u003c/sup\u003e. Faced with this ambiguity, DACs are forced to make subjective judgments about whether requests for data access truly align with permitted data uses, if they have been preserved following many data exchanges or exist at all. Inconsistencies in how data use terms are articulated in consent forms and subsequently interpreted and executed by DACs across the biomedical ecosystem\u003csup\u003e12\u003c/sup\u003e can lead to delayed and inconsistent data access decisions, and risk violating the terms by which patients or participants contributed their data in the first place.\u003c/p\u003e \u003cp\u003eOther steps in the data access pipeline can also contribute to research delays. Emerging research suggests there is growing inefficiency, inconsistency, and error in the manual, entirely human-mediated review of data access agreements\u003csup\u003e11,13\u003c/sup\u003e which are executed in finalizing approved data access requests. Many researchers furthermore still rely on the traditional method of copying-and-downloading data once approved. The copy-download approach multiplies security risks\u003csup\u003e9\u003c/sup\u003e, and is quickly becoming unreasonable given the expanding size and complexities of genomic datasets\u003csup\u003e14,15\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eStandards\u0026rsquo; developers and software engineers have therefore sought to semi-automate three axes of data access control within cloud environments \u0026ndash; user authentication, review of access requests, and concordance of the proposed research with the data use terms of the data requested\u003csup\u003e12\u003c/sup\u003e. Automated decision support (ADS) systems are a coordinated system of algorithms, software, and ontologies\u003csup\u003e16\u003c/sup\u003e that aid in categorizing, archiving, and/or acting on decision tasks for data access review. The Data Use Oversight System (DUOS) typifies one such automated decision support\u003csup\u003e17\u003c/sup\u003e. In recent beta tests, DUOS was successfully shown to concur 100% of the time with human-decided access requests \u003csup\u003e13\u003c/sup\u003e, and also codifies 93% of genomic datasets in NIH\u0026rsquo;s dbGaP\u003csup\u003e18\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eWhile ADS can supplement human-mediated DACs with semi-automated technical solutions, no systematic investigation has sought to characterize relevant barriers and facilitators to ADS in practice\u003csup\u003e19\u003c/sup\u003e. Moreover, we lack understanding of how DAC members perceive the value added by ADS, if any, on the quality and effectiveness of data access review decisions, as well as what challenges they anticipate in adopting ADS taking into account the myriad organizational structures within which DACs operate.\u003c/p\u003e \u003cp\u003eNow is an opportune time to study the implementation barriers and facilitators to using ADS solutions for data access as their development is converging with large-scale data migration to the cloud and hold the promise of near-instant data access decisions. The genomics community can learn important lessons from previous attempts at (premature) ADS implementation without purposeful stakeholder engagement in public health\u003csup\u003e20\u003c/sup\u003e, law enforcement\u003csup\u003e21\u003c/sup\u003e and in clinical care\u003csup\u003e22\u003c/sup\u003e. In this article, we report empirical findings on the \u0026ldquo;constellation of processes\u0026rdquo; relevant for implementing ADS for genomic data access management and provide practical recommendations for institutional data stewards that are considering or have already implemented ADS in this context.\u003c/p\u003e"},{"header":"2. Methods","content":"\u003cp\u003eWe conducted a qualitative description study that engaged prospective end users of ADS for genomic data governance to explore \u003cem\u003eWhat are the barriers and opportunities of implementing automated workflows to manage access requests to genomic data collections, and what effect do ADS have on DAC review quality and effectiveness?\u003c/em\u003e We adopted Damshroder and colleagues\u0026rsquo; definition of implementation as the \u0026ldquo;critical gateway between an organizational decision to adopt an intervention and the routine use of that intervention\u0026rdquo;\u003csup\u003e23\u003c/sup\u003e in order \u0026ldquo;study the constellation of processes intended to get an intervention into use within an organization\u0026rdquo;\u003csup\u003e23\u003c/sup\u003e. We applied the Consolidated Framework for Implementation Research (CFIR) to compare genomic data access processes and procedures across to better understand implementation processes for automated workflows to manage genomic data access across international, publicly funded genomic data repositories. The CFIR provides a \u0026ldquo;menu of constructs\u0026rdquo; associated with five domains of effective implantation which have been rigorously meta-theorized\u0026mdash;that is, synthesized from many implementation theories (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). In addition, the CFIR provides a practical guide to systematically assess potential barriers and facilitators ahead of an innovation\u0026rsquo;s implementation (L. Damschroder et al. 2015). The CFIR is also easily customizable to unveiling bioethical issues during implementation in genomics and has been applied in prior work (Burke and Korngiebel, 2015; Smit et al., 2020).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe interview guide was developed specifically for this study and is available in \u003cb\u003eSupplementary Materials 2.\u003c/b\u003e\u003c/p\u003e \u003cp\u003e \u003cem\u003e\u0026lt;\u0026lt;\u003c/em\u003eFigure \u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e\u003cem\u003e\u0026gt;\u0026gt;\u003c/em\u003e\u003c/p\u003e"},{"header":"3. Data collection","content":"\u003cp\u003eWe conducted a total of 13 semi-structured interviews with 17 DAC members between 27 April and 24 August 2022. Prospective interviewees indicated their interest in being invited to a follow up interview following their participation in survey published elsewhere\u003csup\u003e24\u003c/sup\u003e. All interviews were conducted virtually and audio/video recorded on Zoom. We used validated interview guides from the official CFIR instrument repository (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://cfirguide.org/evaluation-design/qualitative-data/\u003c/span\u003e\u003cspan address=\"https://cfirguide.org/evaluation-design/qualitative-data/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) to probe the barriers and opportunities of implementing ADS solutions for DAC review of data access requests. Interviewees were also recruited from the Data Access Committee Review Standards Working Group (DACReS WG) chaired by authors VR, JL, and ESD, as well as from an internet search of publicly funded genomic data repositories worldwide.\u003c/p\u003e"},{"header":"4. Data analysis","content":"\u003cp\u003eWe first applied a deductive coding frame to the interview transcripts based on a framework analysis approach (Pope, Ziebland, and Mays 2000) and the publicly accessible CFIR codebook available in the \u003cb\u003eSupplemental Materials 1\u003c/b\u003e. To ensure the reliability of conclusions drawn, two independent reviewers (VR and JB) tested the coding schema on three transcripts until reaching a recommended interrater reliability score of 0.83 before analyzing the remaining qualitative dataset. All coding discrepancies during the coding pilot were resolved by consensus discussion.\u003c/p\u003e \u003cp\u003e \u003cem\u003e\u0026lt;\u0026lt;\u003c/em\u003eTable\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e\u003cem\u003e\u0026gt;\u0026gt;\u003c/em\u003e\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eCFIR Code Application Results\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHigh frequency\u003c/p\u003e \u003cp\u003e\u0026gt;=25\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMedium frequency\u003c/p\u003e \u003cp\u003e10\u0026ndash;25\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eLow frequency\u003c/p\u003e \u003cp\u003e\u0026lt;\u0026thinsp;10\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"3\" nameend=\"c3\" namest=\"c1\"\u003e \u003cp\u003eCode (number of code applications)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTension for Change (98), Relative Advantage (72), Knowledge \u0026amp;Beliefs about the Innovation (47), Structural Characteristics (36), Planning (33), Cosmopolitanism (32), External Policy \u0026amp; Incentives (30), DAC tools (\u003cem\u003ecode created by the team)\u003c/em\u003e (30), Compatibility (30), Needs \u0026amp; Resources of those Served by the Organization (27), Key Stakeholders (25)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCulture (23), Networks \u0026amp; Communications (22), Relative Priority (21), Cost (21), Adaptability (21), Innovation Source (20), Reflecting \u0026amp; Evaluating (19), External Change Agents (18), Formally Appointed Internal Implementation Leaders (17),\u003c/p\u003e \u003cp\u003eAvailable Resources (16), Individual Identification with Organization (15), Access to Knowledge \u0026amp; Information (14), Evidence Strength \u0026amp; Quality (14), Peer Pressure (12), Other Personal Attributes (12), Opinion Leaders (11)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eIndividual Stage of Change (9), Goals \u0026amp; Feedback (9), Champions (9), Implementation Climate (8), Engaging (8), Self-efficacy (7), Leadership Engagement (7), Complexity (6), Trialability (6), Learning Climate (6), Characteristics of Individuals (3), Innovation Participants (3), Readiness for Implementation (2), Executing (2), Design Quality \u0026amp; Packaging (1), Process (1)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e"},{"header":"5. Results","content":"\u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eGeographical, Institutional, and Demographic Background of Participants\u003c/b\u003e \u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003cp\u003eForty-one percent of interviewees worked within U.S.-based DACs, while the remaining 59% of interviewees represented DACs at institutions in Canada, the U.K., Spain, Tunisia, Australia, and Japan (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). Nearly 60% of interviewees worked at a non-profit research institute, 24% represented an academic-affiliated research institution, 12% represented a government research agency, and 6% were affiliated with a research consortium. Seventy-six percent of interview participants identified as female, and 24% as male.\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003e \u003cb\u003ea. Opportunities for ADS\u003c/b\u003e \u003c/p\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eWe categorized the frequency of CFIR implementation factors referenced in our interviews in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e. Our findings suggest that there are three major facilitators to implementing ADS for genomic data governance: 1) external policy and need for efficient workflows, 2) institutional ability to scale the ADS, and 3) interoperability.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eParticipant demographics\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"2\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDAC location\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTotal % (n)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUnited States\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e54 (7)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCanada\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e12 (2)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUnited Kingdom\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e6 (1)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSpain\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e6 (1)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTunisia\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e6 (1)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAustralia\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e18 (3)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eJapan\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e12 (2)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e \u003cp\u003e\u003cb\u003eInstitutional type\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNon-profit Research Institute\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e59 (10)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e\n\u003ch3\u003e1) External policy and need for efficient workflows\u003c/h3\u003e\n\u003cp\u003eParticipants considered adopting ADS to comply with new data sharing mandates from research funders (e.g. National Institutes of Health) and those imposed by peer reviewed journals. The demand for and scope of compliant data access review has had a ripple effect on ethics oversight bodies\u003csup\u003e25\u003c/sup\u003e, including DACs, as a result of these new requirements\u003csup\u003e24\u003c/sup\u003e. Most DAC members we engaged currently perform their reviews manually. Members review all data access requests individually or as a committee and make decisions on each request received in the order they were received. Given the anticipated increase in the number of data access requests\u003csup\u003e26\u003c/sup\u003e, our participants noted the reduced workload and costs associated with ADS could contribute to better review efficiencies, without a concomitant loss in review quality and risk of noncompliance with data use conditions.\u003c/p\u003e \u003cp\u003eWe found that participants perceived ADS could reduce DAC member workload by streamlining the intake process for data access requests and verifying that the request matched the terms of use in the original consent obtained at data collection. Indeed, participants noted the initial screening of Data Access Requests (DARs) was a common rate-limiting step. DACs often begin the review process by verifying that all necessary information is documented in the request (e.g. study purpose, datasets requested, ethics review). This step can be time-consuming because the requirements can vary depending on the researcher\u0026rsquo;s institution and the datasets they request. We requested that participants share a copy of their DAR form before, during, or after the interview to compare what information DACs typically required to process a DAR. We found the form fields as well as length of the DAR (from 3 to 18 pages) differed considerably from DAC to DAC. Our participants believed that this is where ADS could be useful by automatically flagging missing information and documents, verifying the authenticity of a requester\u0026rsquo;s identity and the submitted documents, and then sending notifications to requesters if more information is needed. As one interviewee put it:\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003e\u003cem\u003eBecause one of the biggest concerns in our DAC is that sometimes it takes too much time to be read by all the nine members. \u0026hellip; They\u0026rsquo;re institutional directors or university professors. So I think it will help. Maybe if you have 50% of the work done by an automated system, so you just have to do the 50%. I think \u0026hellip; this will be a good motivation for them saying \u0026lsquo;OK\u0026rsquo; [to implement ADS].\u003c/em\u003e\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003e \u003cem\u003eParticipant M\u003c/em\u003e \u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e\n\u003ch3\u003e2) Scalability\u003c/h3\u003e\n\u003cp\u003eParticipants also believed ADS-enabled workflows could be scalable, cost-effective solutions to management of not just newly generated data, but also for legacy data when grant funding ends because ADS can easily store and quickly present data use conditions and audit past DAC reviews:\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003eActually there are lots of costs related to data sharing, particularly if I'm sharing data from the 1990s, for example. I don't have any money or budget anymore to prepare the data [for secondary uses]. \u0026hellip; And similarly, when it comes to these reports [on data sharing activities], there's no extra money for doing the work to create those reports. But we're having to report back over assets from years, decades in fact. And there was always just a little bit of a hint \u0026lsquo;oh well, maybe we'll find some money\u0026rsquo;. No, no, you have to find it out on your own.\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003e \u003cem\u003eParticipant F\u003c/em\u003e \u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eSome participants also shared that ADS could be helpful when DAC members or data generators leave the institution, disrupting review continuity consistency. Unlike for large, well-funded government repositories, many DACs at smaller institutions lack resources for long-term data preservation and access management for data of increasing complexity and volume:\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003eAs the program scales, the participant diversity scales, the data diversity scales. I think it is almost impossible to see a scenario where we do not rely on some level of automation to support human decision making about what is responsible use.\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003e \u003cem\u003eParticipant J\u003c/em\u003e \u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eI mean potentially as we grow over the years, you know what's going to happen. \u0026hellip; we've also discussed some scenarios, where, for example, we find ourselves with a larger amount of requests coming in, [and] we only accept applications up to certain days and then, we open this next quarter, close it again. But there potentially could be room for automation depending on the increase in request in the coming years.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003e \u003cem\u003eParticipant A\u003c/em\u003e \u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e\n\u003ch3\u003e3) Interoperability\u003c/h3\u003e\n\u003cp\u003eAccording to DAC members we interviews, ADS tools could provide centralized, interoperable solutions to facilitate inter-organizational and international data sharing. Participants perceived that ADS could motivate use of standardized request forms, access agreements, dataset identifiers, and methods for verifying researcher identities:\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003e \u003cem\u003e\u0026ldquo;But this [ADS] will free up a lot of time in the process is it also potentially means that it will become easier for, if you're working in a team to hand off tasks as well because you will have a single system. \u0026hellip; Also, consistency between organizations. If we have multiple organizations take this up, it's going to mean less lead time. [Let\u0026rsquo;s] say people take a new job in a new place. We'll actually have some software that people will recognize and be able to use and uptake, which we've been trying to go towards without ethics approval processes within the hospital and health services\u0026hellip; [standardized] systems makes it easier for actual communication between organizations on processes, because everyone kind of begins to know what's happening.\u003c/em\u003e \u003c/p\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003e \u003cem\u003eParticipant E\u003c/em\u003e \u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003e \u003cb\u003eb. Barriers to implementing ADS\u003c/b\u003e \u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003cp\u003eDespite clear advantages of ADS for genomic data access management, our interviewees identified significant barriers to implementation within DAC workflows, including: 1) lower priority compared to more immediate governance challenges, 2) ill equipped personnel and structures within the institution, 3) costs, and 4) degree of human oversight.\u003c/p\u003e\n\u003ch3\u003e1) Low priority\u003c/h3\u003e\n\u003cp\u003eMany participants reported that institutional leadership prioritized other competing research data needs over investing in new data governance structures (e.g. generating quality data, increasing diversity in datasets, collaborating with underrepresented groups of researchers and participants, and releasing datasets). Participants believed researchers in general understand why quality and effective review of data access is important for responsible genomic data sharing but are firstly concerned with data quality. Another suspected reason that ADS implementation ranked lower on institutional priorities was that there had not yet been a significant data incident. As one participant put it:\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003eI don\u0026rsquo;t think that the program thinks it is a very high priority to streamline any of the [data access oversight] process. I think that it will either take something bad happening and then realizing that we need additional capacities on [DAC], or some other hiccup to really promote that need.\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003e \u003cem\u003eParticipant O\u003c/em\u003e \u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eBecause budgets for data governance are not always included in grants, researchers may be less motivated to invest in the additional, largely unpaid work related to data governance. Insufficient resourcing for data sharing and governance mechanisms prospectively in research study design inevitably challenge the downstream execution of data governance upon deposit of the research data once generated, according to at least one DAC member we interviewed:\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003eWe found that some people don\u0026rsquo;t prioritize [data governance] because it\u0026rsquo;s not helpful to them, because it\u0026rsquo;s not our primary function as a department. You know, we\u0026rsquo;re producing new data. That\u0026rsquo;s usually what people, researchers are doing. They\u0026rsquo;re not thinking about what happens to their old data. So, it\u0026rsquo;s not much of a priority. Having said that, research funders are getting very keen for us to use their data. So, there is that sort of tug [of war]. \u0026hellip; If I go into a senior team meeting, you know, something else will be the priority.\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003e \u003cem\u003eParticipant F\u003c/em\u003e \u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e\n\u003ch3\u003e2) Structural characteristics of an organization\u003c/h3\u003e\n\u003cp\u003eWe also found a close correlation between several structural characteristics of the institution (e.g. years in operation, number of personnel, and database size) and participants\u0026rsquo; perceived barriers to ADS implementation. For instance, many participants served on DACs that were established within the last 1\u0026ndash;3 years coinciding with the creation of the institution\u0026rsquo;s database. As the datasets grow, and more researchers are attracted to the resource, there is greater potential for existing management processes to be overwhelmed. It is precisely at this early juncture that DACs weight their options for ADS tools, and proactively address relevant barriers ahead of future plans for implementation. Some DAC members preferred to gain more experience with existing data access management in these early years of data release before integrating ADS \u003cem\u003e\u0026ldquo;because we\u0026rsquo;re not sure how [name of participant\u0026rsquo;s country] citizens feel or consider about the automatic decision on data sharing.\u0026rdquo; Participant K.\u003c/em\u003e\u003c/p\u003e\n\u003ch3\u003e3) Cost\u003c/h3\u003e\n\u003cp\u003eWhile cost was not a primary concern for ADS implementation at well-funded big data repositories, it was a significant barrier for DAC members working at smaller repositories, individual research departments, or research programs associated with a genomics consortium who were more often supported by research grants or contracts rather than an independent funding source.\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003e \u003cem\u003e\u0026ldquo;We [data governance office] are supported through project-specific funding. \u0026hellip; Governance ends up being a little bit of this indirectly supported component of our work and services. That has limited the ways in which we can innovate around governance. \u0026hellip; We don\u0026rsquo;t have a huge budget.\u0026rdquo;\u003c/em\u003e \u003c/p\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003e \u003cem\u003eParticipant N\u003c/em\u003e \u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eWithout dedicated budget for human and material resources, some DAC members were concerned that the initial investment in ADS and significant changes to current workflows would be key issues, to say nothing of new education and training materials and updates to internal policies, among other ancillary revisions to internal workflows.\u003c/p\u003e\n\u003ch3\u003e4) Lack of human oversight\u003c/h3\u003e\n\u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eWhile some DAC members were enthusiastic about improvements in efficiency and consistency of ADS, participants unanimously rejected the idea of fully automating access management: \u003cem\u003e\u0026ldquo;no matter what we do with automation that I feel there always needs to be that human element who's coming in and checking. So, there will always be that barrier to upscaling\u0026rdquo; Participant E.\u003c/em\u003e Other participants emphasized that prior to implementation, they would need to gauge how research participants at their own institution as well as the general public would react to ADS for data access review.\u003c/p\u003e \u003cp\u003eParticipants were also skeptical that ADS could adequately assess complex, sensitive data reuse issues which they felt required a deep understanding of ethical, legal, and sociocultural contexts within which data were collected, used, and shared. Some DAC members reported asking data requesters to clarify their study purpose and justify their need for specific datasets in recognition of these sociocultural dimensions.\u003c/p\u003e \u003cp\u003e \u003cem\u003eI\u0026rsquo;m also someone who thinks that it's important to be very critical about what's the nature of the work being done. Maybe it's solid from a scientific point of view. But are there other concerns from other perspectives that need to be taken into account? That is partly why we have community members on the [committee], and that's something I'm not sure can be simplified or automated.\u0026rdquo;\u003c/em\u003e \u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003e \u003cem\u003eParticipant B\u003c/em\u003e \u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eHowever, when it comes to automating anything that requires reviewing information where there might be a lot of nuances, where there might be a lot of interpretation that's required, I'm a little bit more hesitant simply because I think to some extent you do need some room for a little bit of mulling over the information, \u0026hellip; and I think there are some information that come through with requests, that don't neatly fit into check boxes.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003e \u003cem\u003eParticipant B\u003c/em\u003e \u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e"},{"header":"6. Discussion","content":"\u003cp\u003eParticipants overall perceived that ADS tools could be well positioned to help DACs streamline data access compliance. While believed to beneficial, ADS solutions were unlikely to immediately or directly advance the research organization\u0026rsquo;s core mission (e.g. collecting quality data and driving scientific discoveries and innovations) according to the DAC members we engaged. Participants also tended to regard ADS implementation, as well of data governance workflow solutions as a comparatively low priority compared to regulatory compliance, investigator support, and database curation, among other competing demands on DAC member time. Most research grants allow investigators to apply for support for data collection and analysis, but rarely to establish actual governance structures needed to stand up access management services. DAC members therefore shared that it can be difficult to convince institutional leadership to invest in data governance solutions such as ADS. Some participants noted that the real benefit of ADS is to streamline workflows that are currently time consuming and tedious, for example preliminary review of applications to ensure there are no missing elements. Applying ADS to facilitate these workflows could free DAC members to dedicate more time to deliberate on more substantive ethics issues raised by data access requests.\u003c/p\u003e \u003cp\u003eWhile data governance has often been considered auxiliary work, new research findings and new U.S. federal government policies, such as the National Institutes of Health Data Management and Sharing policy (DMS), have elevated its importance by placing additional requirements for data sharing \u003csup\u003e27\u003c/sup\u003e. Some of our study participants mentioned that they believe researchers are drawn to their databases not because of their data access policies and practices, but because of the quality and diversity of their datasets. This perspective differs from what another study found suggesting that ease of access is at least marginally important when choosing databases/institutions for research data\u003csup\u003e28\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eThe DMS policy will accelerate the accumulation of an enormous number of datasets, which will, absent interventions, significantly push up the costs associated with data storage and management. If not developed in collaboration, institutional databases created for the explicit purpose of sharing research data generated from federal funds has the potential to introduce myriad access pathways that make the data effectively \u0026lsquo;shared\u0026rsquo; but unfindable. One of our participants mentioned that some legacy data, which date back to the 50s and 60s, were transferred to their organization due to a closure of the repositories which originally stored these datasets. This case suggests that there is an urgent need for more efficient and sustainable solutions for data access management and sharing and perhaps is reasonable cause to ensure that there is a contingency plan for publicly-funded data shared via non-publicly supported repositories in the event the repository closes or changes in policy or personnel.\u003c/p\u003e \u003cp\u003e \u003cb\u003eLimitations\u003c/b\u003e \u003c/p\u003e \u003cp\u003eOur results should be considered in light of several methodological limitations. While geographically diverse, many of our interview participants were affiliated with DACs based at large, well-resourced research institutions. It is likely that responses and perceptions of implementation factors related to ADS would differ substantially if more DACs from low- or under-resourced institutions were represented in our sample. Our data collection design relies on self-reports of institutional data access policy and procedures. Many interview participants were aware of the Global Alliance for Genomics and Health, and the data access committee review standards we were principally involved in developing\u003csup\u003e29\u003c/sup\u003e. Thus while we endeavored to create a safe, open environment for participants to share their honest views, social desirability bias related to our prior work may have influenced how participants responded. Lastly, CFIR predefines sociological constructs relevant to implementation. Our analysis was therefore limited only those constructs covered in the framework when others might have emerged inductively if we adopted an alternative analytic frame.\u003c/p\u003e "},{"header":"Conclusion","content":"\u003cp\u003eIn this article, we reported findings from semi-structured qualitative interviews with DAC members from around the world on the relevant barriers and facilitators of implementing ADS for genomic data access management. Our findings suggest there is general support for pilot studies that test ADS performance in certain elements of the data access management workflows, such as cataloging data types, verifying user credentials and tagging datasets for use terms. Participants indicated that ADS should supplement, but not replace DAC member work. This sentiment was especially strong with respect to tasks that were perceived to require sensitivity and human value-judgments such as privacy protections, group harms, and study purpose. Rather, our findings offer cautious optimism regarding the ways in which algorithms, software, and other machine-readable ontologies can streamline aspects of DAC decision-making while also enabling new opportunities for improving consistency and fairness in DAC decisions.\u003c/p\u003e \u003cp\u003eTo that end, we conclude with practical recommendations for institutional data stewards that are considering or have already implemented ADS for data access management. The explosion in the volume and complexity of genomic and associated health data is converging with the need to manage access more efficiently to these data. Such trends point intuitively to solutions that can help to alleviate, or at least prevent bottlenecks in the access process to preserve the scientific and social value of data generated from public investments in research.\u003c/p\u003e \u003cp\u003eTo put ADS solutions to the test, future research should compare access decisions and their outcomes between institutions who do/not use such tools for data access management; whether ADS delivers on its efficiency promises and liberates DAC member time previously spent addressing procedural matters and instead create more opportunities for committee deliberation on substantive ethics issues.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors wish to thank members of the Data Access Committee Review Standards Working Group, and the Regulatory and Ethics Work Stream of the Global Alliance for Genomics and Health for their contributions to the intellectual community that inspired this work. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor Contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAuthors VR, JL and ED conceptualized, designed, and carried out the study. Author JB led in the data collection and analysis and drafting of early manuscript drafts. All authors, VR, JB, JL and ED took part in writing and editing the manuscript, responding to peer reviewer comments \u0026nbsp;and approved the final version.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study was reviewed and approved the Stanford University Institutional Review Board. All participants were informed of the purpose of the study, funding, risks and benefits at the time of invitation. A research assistant consented all participants prior to the interview, and were provided opportunities to ask any questions.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting Interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll authors are members of the Global Alliance for Genomics and Health. JL is lead software engineer of the Data Use Ontology and is a member of the Broad Institute Data Access Committee.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eMaterials described in the manuscript and data supporting our findings can be made available upon request. All requests should be directed to Vasiliki Rahimzadeh, PhD at [email protected].\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eStephens ZD, Lee SY, Faghri F, Campbell RH, Zhai C, Efron MJ, Iyer R, Schatz MC, Sinha S, Robinson GE. Big Data: Astronomical or Genomical? PLoS Biol. 2015;13:e1002195. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1371/journal.pbio.1002195\u003c/span\u003e\u003cspan address=\"10.1371/journal.pbio.1002195\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eIQVIA Institute for Human Data. Science Understanding the Global Landscape of Genomic Initiatives: Progress and Promise.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCook-Deegan R, McGuire AL. Moving beyond Bermuda: Sharing data to build a medical information commons. Genome Res. 2017;27:897\u0026ndash;901. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1101/gr.216911.116\u003c/span\u003e\u003cspan address=\"10.1101/gr.216911.116\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEvans B. (2019). Genomic data commons. Governing Privacy in Knowledge Commons, 75\u0026ndash;105.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eContreras JL, Knoppers BM. The genomic commons. Annu Rev Genom Hum Genet. 2018;19:429\u0026ndash;53. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1146/annurev-genom-083117-021552\u003c/span\u003e\u003cspan address=\"10.1146/annurev-genom-083117-021552\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGrossman RL. Ten lessons for data sharing with a data commons. Sci Data. 2023;10:120. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41597-023-02029-x\u003c/span\u003e\u003cspan address=\"10.1038/s41597-023-02029-x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCheah PY, Piasecki J. Data Access Committees. BMC Med Ethics. 2020;21:1\u0026ndash;8. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12910-020-0453-z\u003c/span\u003e\u003cspan address=\"10.1186/s12910-020-0453-z\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShabani M, Thorogood A, Borry P. Eur J Hum Genet. 2016;24:1671\u0026ndash;5. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/ejhg.2016.111\u003c/span\u003e\u003cspan address=\"10.1038/ejhg.2016.111\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. Who should have access to genomic data and how should they be held accountable? Perspectives of Data Access Committee members and experts.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRamos EM, Din-Lovinescu C, Bookman EB, McNeil LJ, Baker CC, Godynskiy G, Harris EL, Lehner T, McKeon C, Moss J, GAIN Data Access Committee. Am J Hum Genet. 2013;92:479\u0026ndash;88. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.ajhg.2012.08.034\u003c/span\u003e\u003cspan address=\"10.1016/j.ajhg.2012.08.034\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. A Mechanism for Controlled Access to GWAS Data: Experience of the.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShabani M, Knoppers BM, Borry P. EMBO Mol Med. 2015;7:507\u0026ndash;9. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.15252/emmm.201405002\u003c/span\u003e\u003cspan address=\"10.15252/emmm.201405002\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. From the principles of genomic data sharing to the practices of data access committees.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMello MM, Triantis G, Stanton R, Blumenkranz E, Studdert DM. Waiting for data: Barriers to executing data use agreements. Science. 2020;367:150\u0026ndash;2. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1126/science.aaz7028\u003c/span\u003e\u003cspan address=\"10.1126/science.aaz7028\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLawson J, Cabili MN, Kerry G, Boughtwood T, Thorogood A, Alper P, Bowers SR, Boyles RR, Brookes AJ, Brush M, et al. The Data Use Ontology to streamline responsible access to human biomedical datasets. Cell Genomics. 2021;1:100028. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.xgen.2021.100028\u003c/span\u003e\u003cspan address=\"10.1016/j.xgen.2021.100028\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLawson J. Empirical Validation of an Automated Approach to Data Use Oversight. Cell Genomics Forthcoming; 2021.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVan der Auwera G, O\u0026rsquo;Connor B. Genomics in the Cloud: Using Docker, GATK, and WDL in Terra. 1st ed. O\u0026rsquo;Reilly Media; 2020.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBirger C, Hanna M, Salinas E, Neff J, Saksena G, Livitz D, Rosebrock D, Stewart C, Leshchiner I, Baumann A, et al. FireCloud, a scalable cloud-based platform for collaborative genome analysis: Strategies for reducing and controlling costs. (Bioinformatics). 2017. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1101/209494\u003c/span\u003e\u003cspan address=\"10.1101/209494\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRahimzadeh V, Lawson J, Rushton G, Dove ES. Leveraging Algorithms to Improve Decision-Making Workflows for Genomic Data Access and Management. Biopreserv Biobank. 2022. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1089/bio.2022.0042\u003c/span\u003e\u003cspan address=\"10.1089/bio.2022.0042\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eData Use Ontology. (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLawson J, Ghanaim EM, Baek J, Lee H, Rehm HL. Aligning NIH\u0026rsquo;s existing data use restrictions to the GA4GH DUO standard. Cell Genomics. 2023;3:100381. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.xgen.2023.100381\u003c/span\u003e\u003cspan address=\"10.1016/j.xgen.2023.100381\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRahimzadeh V, Lawson J, Baek J, Dove ES. Automating Justice: An Ethical Responsibility of Computational Bioethics. Am J Bioeth. 2022;22:30\u0026ndash;3. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/15265161.2022.2075051\u003c/span\u003e\u003cspan address=\"10.1080/15265161.2022.2075051\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eObermeyer Z, Powers B, Vogeli C, Mullainathan S. Dissecting racial bias in an algorithm used to manage the health of populations. Science. 2019;366:447\u0026ndash;53. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1126/science.aax2342\u003c/span\u003e\u003cspan address=\"10.1126/science.aax2342\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHartzog W, Conti G, Nelson J, Shay LA. Inefficiently Automated Law Enforcement. Mich State Law Rev. 2016;1763:1763\u0026ndash;96.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLedford H. Millions affected by racial bias in healthcare algorithm. Nature. 2019;574:608\u0026ndash;10.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDamschroder LJ, Aron DC, Keith RE, Kirsh SR, Alexander JA, Lowery JC. Fostering implementation of health services research findings into practice: a consolidated framework for advancing implementation science. Implement Sci. 2009;4. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/1748-5908-4-50\u003c/span\u003e\u003cspan address=\"10.1186/1748-5908-4-50\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLawson J, Rahimzadeh V, Baek J, Dove ES. (2023). Achieving Procedural Parity in Managing Access to Genomic and Related Health Data: A Global Survey of Data Access Committee Members. Biopreservation and Biobanking, bio.2022.0205. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1089/bio.2022.0205\u003c/span\u003e\u003cspan address=\"10.1089/bio.2022.0205\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRahimzadeh V, Serpico K, Gelinas L. Institutional review boards need new skills to review data sharing and management plans. Nat Med. 2023;1\u0026ndash;3. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41591-023-02292-w\u003c/span\u003e\u003cspan address=\"10.1038/s41591-023-02292-w\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCabili MN, Lawson J, Saltzman A, Rushton G, O\u0026rsquo;Rourke P, Wilbanks J, Rodriguez LL, Nyronen T, Courtot M, Donnelly S, et al. Empirical validation of an automated approach to data use oversight. Cell Genomics. 2021;1:100031. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.xgen.2021.100031\u003c/span\u003e\u003cspan address=\"10.1016/j.xgen.2021.100031\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFinal NIH. Policy for Data Management and Sharing \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://grants.nih.gov/grants/guide/notice-files/NOT-OD-21-013.html\u003c/span\u003e\u003cspan address=\"https://grants.nih.gov/grants/guide/notice-files/NOT-OD-21-013.html\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTrinidad MG, Ryan KA, Krenz CD, Roberts JS, McGuire AL, De Vries R, Zikmund-Fisher BJ, Kardia S, Marsh E, Forman J, et al. Extremely slow and capricious: A qualitative exploration of genetic researcher priorities in selecting shared data resources. Genet Sci. 2023;25:115\u0026ndash;24. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.gim.2022.09.003\u003c/span\u003e\u003cspan address=\"10.1016/j.gim.2022.09.003\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGlobal Alliance for Genomics and Health. (2021). Data Access Committee Guiding Principles and Procedural Standards Policy.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-medical-ethics","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"meth","sideBox":"Learn more about [BMC Medical Ethics](http://bmcmedethics.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/meth/default.aspx","title":"BMC Medical Ethics","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-3849259/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-3849259/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eData access committees (DAC) gatekeep access to secured genomic and related health datasets yet are challenged to keep pace with the rising volume and complexity of data generation. Automated decision support (ADS) systems have been shown to support consistency, compliance, and coordination of genomic data sharing of data access review decisions. However we lack understanding of how DAC members perceive the value add of ADS, if any, on the quality and effectiveness of their reviews. In this qualitative study, we report findings from 13 semi-structured interviews with DAC members from around the world to identify relevant barriers and facilitators to implementing ADS for genomic data access management. Participants generally supported pilot studies that test ADS performance for example in cataloging data types, verifying user credentials and tagging datasets for use terms. Concerns related to over-automation, lack of human oversight, low prioritization, and misalignment with institutional missions tempered enthusiasm for ADS among the DAC members we engaged. Tensions for change in institutional settings within which DACs operated was a powerful motivator for why DAC members considered the implementation of ADS into their access workflows, as well as perceptions of the relative advantage of ADS over the status quo. Future research is needed to build the evidence base around the comparative effectiveness and decisional outcomes of institutions that do/not use ADS into their workflows.\u003c/p\u003e","manuscriptTitle":"A qualitative interview study to determine barriers and facilitators of implementing automated decision support tools for genomic data access","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-01-17 21:14:05","doi":"10.21203/rs.3.rs-3849259/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2024-04-01T07:29:01+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2024-03-29T13:07:26+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2024-02-02T15:10:02+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"9a6f4d82-a951-46d3-85a8-79b5ef93efb0","date":"2024-01-22T14:03:56+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2024-01-22T13:29:44+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2024-01-16T07:10:37+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2024-01-16T07:09:07+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Medical Ethics","date":"2024-01-09T22:08:01+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-medical-ethics","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"meth","sideBox":"Learn more about [BMC Medical Ethics](http://bmcmedethics.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/meth/default.aspx","title":"BMC Medical Ethics","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"7b871a2d-3650-45dd-9329-ec3ad18663d0","owner":[],"postedDate":"January 17th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2024-05-05T12:18:03+00:00","versionOfRecord":{"articleIdentity":"rs-3849259","link":"https://doi.org/10.1186/s12910-024-01050-y","journal":{"identity":"bmc-medical-ethics","isVorOnly":false,"title":"BMC Medical Ethics"},"publishedOn":"2024-05-05 12:18:03","publishedOnDateReadable":"May 5th, 2024"},"versionCreatedAt":"2024-01-17 21:14:05","video":"","vorDoi":"10.1186/s12910-024-01050-y","vorDoiUrl":"https://doi.org/10.1186/s12910-024-01050-y","workflowStages":[]},"version":"v1","identity":"rs-3849259","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-3849259","identity":"rs-3849259","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}