ALLA DIGILOVA, REBECCA LI, MARK BARNES, AND BARBARA BIERER
APPEALS FOR increased clinical trials data transparency date back at least a decade, but thanks to a series of catalytic events, efforts to broaden access to clinical trials data have accelerated in recent years. The complementary effects of regulatory trends in the United States and Europe, academic journals’ new data disclosure requirements, and voluntary efforts by the industry have led to an unprecedented growth of data-sharing mechanisms within the clinical research enterprise. The resulting access to clinical trials information has important implications for academicians, pharmaceutical companies, other research funders, study participants, and the biomedical sciences. While transparency and data sharing are laudable goals, more open access may portend risks to privacy and commercial interests and may alter the behavior of regulatory agencies. This chapter will outline the features of a well-designed data-sharing model that would allow for the realization of associated benefits while protecting against the underlying risks and will review recent trends toward participant-level clinical trials data sharing.
As a preliminary matter, it is important to clarify that this chapter focuses on the most recent efforts in sharing clinical trials participant-level data sets as distinct from summary level, aggregated, or pooled results. Summary level results do not include a listing of individual level data and outcomes, thereby carrying a lower risk of reidentification but only offering limited utility for secondary research. While summary level data sharing is vital for full disclosure of both positive and negative trial results, data sharing intended to facilitate collaborative advancement of science is better served through access to participant-level data that allows secondary researchers to replicate previous analyses, generate new hypothesis, and conduct novel studies.
I. RISKS, BENEFITS, AND ISSUES OF INCREASED DATA TRANSPARENCY
Participant-level data sharing requires careful consideration of associated risks and benefits and implications for future system design. Participant-level data sharing needs to account for concerns regarding participants’ privacy and the risk of reidentification, public trust, sponsors’ commercial interests, research quality, the financial costs, and overall viability of the clinical research enterprise. Critical evaluation and analysis of modern data-sharing steps should, however, consider and value the underlying positive goals of broader data access—advancement of scientific knowledge and treatment innovation.
The difficulties inherent in broadening access to participant-level clinical trials data cannot be underestimated. Standard research practice, guided by the mandates of the Declaration of Helsinki (WMA 2013) and other applicable regulations, requires researchers to protect study participants’ privacy and confidentiality. In the modern era of information, releasing participant-level data, as compared to pooled, summary level data, increases the probability that study participants may be reidentified; well-publicized studies illustrate the possibility of reidentification through the utilization of publicly available information (Narayanan and Shmatikov 2008). Fears of data misuse and actual instances of participant reidentification, even if rare, could cause significant adverse consequences for the clinical trial enterprise, reducing public trust and further decreasing future participation in clinical trials. However, while the risks of reidentification necessitate robust de-identification standards and methodologies, stripping any data set of too many identifiers, or grouping variables in wide ranges, will render the data less useful for secondary analyses, thus undermining the purpose of data sharing. A carefully balanced approach is, therefore, necessary. Supplementing data-protection safeguards with restrictions on data access, with potential penalties for misuse, is one efficient way to achieve the desired balance.
In addition to managing the risk of patient reidentification, controlling access to participant-level clinical trials data will also accommodate industry concerns regarding confidential information and competitive efforts. A robust data-sharing environment relies on the industry’s willingness to participate. Yet, as exemplified by the June 2013 European Medicines Agency (EMA) proposal that would require industry to make participant-level clinical trials data public, expansive access can engender forceful opposition from industry, which counters that mandatory release of certain documents may interfere with commercial confidentiality and threaten disclosure of trade secrets (EMAa 2013).
Mandatory disclosure, as proposed by the EMA policy, raises the specter of competitive spying. Rival companies may use information contained in new drug applications submitted in one jurisdiction to seek regulatory approval of competing products in another jurisdiction without incurring associated costs (Mello et al. 2013). Competitors may also peruse disclosed data to learn of submitters’ general scientific and commercial strategies, with the goals of informing their development of rival products. These consequences of unmediated, uncurated data sharing may damage profitability of new research and correspondingly decrease industry’s willingness to invest in developing new treatments and medicines (Mello et al. 2013). Companies may also support agendas that are distasteful to industry and perhaps even harmful to public discourse, such as “fishing expeditions” seeking masses of data so that in any way possible evidence can be developed that would dispute treatment efficacy claims. These issues can be addressed by controlled access data-sharing models, providing sponsors the opportunity to redact data of commercially sensitive information. However, if a sponsor is also entrusted with final authority to grant or deny data access requests, the credibility of data-sharing systems will suffer. A neutral body that determines whether and how data can be shared, and the conditions of use, would resolve this tension.
Permitting wider access to participant-level clinical trials data also requires weighing the costs and benefits to research quality. Here, open and uncontrolled access to clinical trials data may result in studies conducted by unqualified researchers who reach erroneous conclusions with unfortunate risks to public health. Rather than achieving the intended goals of advancing knowledge and treatment of disease, poorly performed analyses will confound scientific and public understanding of disease and treatment. Further, already underfunded regulatory agencies will need to reexamine any new data with respect to their earlier conclusions and analyses. Limiting access to data to qualified investigators who have the requisite training, skills, and tools can be one step to assure quality analysis. Assuring scientific quality will foster progress of science and public health by generating new hypotheses, uncovering new relationships, and speeding the process of finding new treatments.
Lastly, broadening access to clinical trials data should take into account globalization of the clinical trials enterprise and competing regulatory requirements of various jurisdictions. Local differences in applicable legal regimes, including trade secrecy/commercial confidentiality laws, privacy regulations, and emerging data disclosure laws, can hinder sharing efforts. For example, if a clinical trial were conducted in a nation that restricts secondary use of data, the sponsor may be unable to file for marketing authorization in the European Union if the proposed EMA policy, mandating disclosure of clinical trials data for future research, takes effect. Development of common national sets of rules and practices is therefore necessary to foster a successful clinical trials data-sharing environment that encompasses the increasingly multinational clinical trials enterprise.
The interplay of risks and benefits associated with access to participant-level data reinforces the need for a well-designed data-sharing regime. The rigor of requirements for data access should balance the need to allow access to qualified researchers with the need to protect against data misuse and poor quality or even erroneous analyses. Data access and denial decisions should be free from competitive bias and real or perceived conflict of interest. Satisfactory standards of de-identification and data-access protection are essential and should be adequate to minimize the risk of identifying study participants while maintaining the utility of the data set for future research.
One example of a data-sharing model that can provide for appropriate management of risks and benefits is a learned intermediary model, consisting of an independent, third-party decision-making authority basing data access decisions on specific standards (Mello et al. 2013). Unlike models that rely on data sponsors to review the requests and/or retain custody of the data, by vesting control over decision making in a neutral party, the learned intermediary model can guard against biased determinations by the data sponsor and thus improve public trust in the system (Mello et al. 2013). At the same time, in contrast to open access models, the learned intermediary model can help prevent flawed analyses and can foster research quality through evaluating and approving proposals based on scientific merit and the requesting party’s expertise.
II. HISTORY AND CATALYSTS OF INCREASED DATA TRANSPARENCY
In light of the risks and benefits described previously, the next section outlines historical trends in clinical trials data sharing and the various approaches undertaken by policy makers, academic journals, and voluntary efforts to broaden access to participant-level data. We pay particular attention to influential initiatives that have served as catalysts for greater transparency in clinical trials data.
A. Efforts by Policy Makers
Over the past decade, policy makers and governmental institutions in the United States and abroad have heard and responded to public demands for greater clinical trials data transparency stemming from mistrust of industry and perceived lack of information that has been made voluntarily available. Over time, regulators have exhibited increasingly greater receptiveness toward data sharing, shifting from more restrictive traditional approaches—demanding trial registration in public databases—to provisions of aggregate and participant-level data. This section will discuss some of the prominent catalysts in the regulatory field responsible for fostering broader access to clinical trials data.
1. Early Registration Requirements
Before the international debate to allow access to participant-level data gained force, the U.S. government undertook steps to provide the public with selected information from clinical trials. In 1997, as part of the Food and Drug Administration Modernization Act, Congress required trial sponsors to register certain clinical trials and directed the National Institutes of Health (NIH) to establish a public database resource of clinical trials conducted under investigational new drug applications (INDs) for drugs intending to treat serious or life-threatening diseases or conditions (NIH 2013a). In February 2000, NIH unveiled ClinicalTrials.gov, an online resource of open clinical trials for experimental treatments that includes trial objectives, eligibility requirements, location, and point of contact; the website was made available to the public and became one of the first trial registries of its kind. Starting in 2009, the ClinicalTrials.gov database was expanded in order to facilitate the required posting of summary results. While originally focused on trial registration and the public availability of information about possible clinical trials in which patients might enroll, ClinicalTrials.gov was the first step toward more comprehensive data sharing. Yet its initial goal was to improve patients’ access to trials rather than to promote secondary research analyses.
In November 2004, the World Health Organization (WHO) identified the need to establish an international voluntary platform to register clinical trials; that platform became the International Clinical Trials Registry Platform (ICTRP). In establishing this resource, WHO underscored the importance of trial registrations for overcoming the negative effects of publication bias and selective reporting on evidence-based health care practices (WHO 2014). In September 2005, the International Committee of Medical Journal Editors (ICMJE) unveiled a new policy requiring trial registration prior to first patient enrollment in a publicly available database as a condition of later publication of study results in any of the ICMJE-affiliated journals. The ICMJE policy statement articulated a perceived obligation of the clinical trials community members to respect the altruism and trust of study participants in dedicating themselves and their time to advance scientific knowledge. According to the policy statement, the obligation demands ethical research conduct and honest reporting. While the ICMJE policy has not mandated more advanced data-sharing requirements, it recognized that “[r]egistration is only part of the means to an end; that end is full transparency with respect to performance and reporting of clinical trials” (De Angelis et al. 2004). ICJME laid the predicate for later, more comprehensive data-sharing initiatives and mechanisms.
Reinforcing these requirements, the 2008 revised Declaration of Helsinki mandated that “[e]very research study involving human subjects must be registered in a publicly accessible database before recruitment of the first subject” (WMA 2013). Trial registration thereby became effectively unavoidable, with registration databases spreading through all parts of the globe, reaching, among others, the European Union, Japan, Australia, and African nations.
2. NIH Sharing Research Data Policy
In March 2003, NIH released its Statement on Sharing Research Data. The policy required all NIH-grant applicants seeking a budget exceeding $500,000 in any single year to include a detailed data-sharing plan. Foreshadowing themes in the current data-sharing debate, the policy sought to balance competing interests of improved clinical trials data access with the need to protect patients’ privacy, emphasizing that data intended for broader use must meet appropriate anonymization standards. The policy provided an important break from earlier transparency initiatives that had primarily focused on improving patients’ access to clinical trials; instead, the new NIH initiative became one of the first concerted efforts to encourage data sharing in order to promote the flow of information among researchers for the benefit of science and public health.
3. Food and Drug Administration Amendments Act (FDAAA) of 2007
The U.S. Congress took significant additional steps to promote availability of clinical trials information with its passage of the Food and Drug Administration Amendments Act (FDAAA) of 2007. Pertinent to clinical trials transparency, FDAAA extended the applicability and scope of ClinicalTrials.gov registration requirements to mandate reporting of summary results and adverse events for certain trials (NIH 2013b). The effect of FDAAA on data sharing was further strengthened by a concurrent NIH policy requiring all future grantees in applicable clinical trials to certify compliance with FDAAA, including its clinical trial registration and results reporting requirements (NIH 2003).
4. Recent Trends Toward Participant-Level Data Sharing
As the possible benefits of access to raw clinical trials data gained broader recognition and voluntary data-sharing mechanisms emerged, in summer 2013, FDA and the European Medicines Agency (EMA) of the European Union introduced policy proposals aimed to allow public access to participant-level clinical trials data submitted by pharmaceutical (and in the case of FDA, medical device and biotechnology) companies to the agencies as part of a drug approvals process. Both the request for comment (FDA) and the proposed policies (EMA) stood in stark contrast to regulators’ historical treatment of industry data as protected against disclosure in order to safeguard commercial interests and prompted a passionate debate among a broad range of stakeholders.
a. FDA’s role in data transparency: FDA has traditionally played a circumscribed role in participant-level data-sharing efforts, based on the position that commercial confidentiality laws prevent voluntary public disclosure of data submitted to the agency (Mello 2013). Following a marketing authorization, FDA generally voluntarily releases only two kinds of clinical trials data received during a new drug approval process: (1) a summary basis of approval with limited description of results of the clinical trials in humans, and (2) a synopsis of reviews, including medical and pharmacological opinions, made by the agency staff during the approval process (Kesselheim and Mello 2007). Additional disclosures may be made through release of FDA advisory committee process reports. Releases are routinely reviewed internally before disclosure and are redacted for sensitive and commercially confidential information (Kesselheim and Mello 2007).
In view of the historically limited voluntary disclosure by FDA, researchers, consumer groups, and advocates have typically relied on the Freedom of Information Act (FOIA) requests to gain access to clinical trials data submitted to the agency (Kesselheim and Mello 2007). However, the utility of FOIA for data sharing is significantly impaired by FDA’s legal responsibilities toward protecting drug manufacturers’ commercially sensitive information, mandated by Exemption 4 to FOIA (5 USC § 552(b)(4)), and FDA’s obligations to guard trial participants’ privacy pursuant to international principles of good clinical practices (FDA 2010). When FDA refuses disclosure pursuant to a FOIA exemption, the only recourse of the requesting party is complicated, time-consuming, and expensive litigation with uncertain likelihood of success (Kesselheim and Mello 2007).
While respecting the mandates of industry confidentiality and participant privacy, FDA recently proposed for public comment some measures to promote greater accessibility to clinical trials data (FDA 2013). In June 2013, FDA published a proposal for public comment on Availability of Masked and De-identified Non-Summary Safety and Efficacy Data. The proposal would permit public release of pooled, de-identified, and “masked” clinical and preclinical data from marketing applications submitted to FDA. “Masked” data was defined by FDA as data removed from any link to a specific drug or device product or application. Though inherently less useful for some secondary research purposes than data identifying specific products, masked data can be used to build disease progression models, characterize natural history of illnesses, assess population responses to standard-of-care products, or characterize risk factors (FDA 2013). It currently remains unclear whether the agency will move forward with the proposal and how the process of data sharing would be structured.
b. European Medicines Agency’s proposed policies: The European Union has signaled changes in its regulatory policies regarding data sharing and has gradually moved from a more restrictive to a more open approach. Since November 2010, in accordance with its new policy on access to documents related to medicinal products for human and veterinary use, the EMA has been releasing upon request certain contents of applications submitted for marketing authorization. Documents available for release include clinical study reports for medicines on which the agency has completed its decision-making process. Further extending its approach to transparency, in June 2013, the EMA published a new draft policy on Publication and Access to Clinical-Trial Data (EMA 2013a). Citing hopes that new policy will enable the “wider scientific community to make use of detailed and high-quality clinical trial data to develop new knowledge in the interest of public health,” the proposal laid out a framework that would permit significantly expanded and unprecedented public access to participant-level clinical trials data.
The EMA’s new approach faced stiff opposition from the drug manufacturing industry, among others. The European Federation of Pharmaceutical Industries and Associations (EFPIA), the trade association of the research-based pharmaceutical industry in Europe, submitted public comments on the draft policy, stressing the risks and likely “negative implications” associated with the EMA’s proposal (EFPIA 2013). In a bid to block disclosure of clinical trials data, two pharmaceutical companies, AbbVie and InterMune, initiated legal actions against the EMA, arguing that data intended for sharing by the agency is commercially confidential information and that publication of such information would violate European Union (EU) privacy laws and the International Agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPS) (AbbVie v. EMA 2013; InterMune v. EMA 2013). By July 2014, however, the EMA had settled the lawsuit filed by AbbVie by agreeing to release a more limited data set, and InterMune has dropped its lawsuit against the agency.
The draft policy sparked an animated public debate. More than 1,000 comments from a diverse group of stakeholders were submitted to the EMA, which postponed final action on its data-sharing proposal until June 2014 (EMA 2013b). In a parallel development, in April 2014, the European Parliament approved a new regulation regarding clinical trials on medical products for human use. Among other things, the new regulation extended trial registration requirements, mandated publication of study results within a year of trial’s end, and required sponsors to make clinical study reports publicly available (EMA 2014).
III. BRITISH MEDICAL JOURNAL POLICY AND VOLUNTARY INDUSTRY AND OTHER EFFORTS
A. British Medical Journal Policy
Clinical trials data-sharing efforts gained significant traction in the years following the 2005 policy statement put forth by ICJME. Individual sponsor data-sharing models, precompetitive consortia, and public–private partnerships were developed for data sharing, each with different and diverse foci and varying degrees of access. For instance, GlaxoSmithKline (GSK), in a project known as “Share,” began to permit controlled access to anonymized patient-level data under a data use agreement that allowed requesters unprecedented access to industry-sponsored clinical trials data (Harrison 2012). As the research community grew more accustomed to the concept of safeguarded clinical trials data access, including access to anonymized patient-level data, the British Medical Journal (BMJ) introduced new, robust data-sharing requirements in October 2012 (Godlee 2012).
Starting in January 2013, the BMJ required authors to commit to sharing, upon “reasonable request,” “all anonymized data on individual patients on which the analysis, results, and conclusions reported in the paper are based” as a condition for publication (Godlee 2012). To ensure process transparency, the policy required all data requesters to make the requests public and submit reanalysis protocols to the primary authors. Authors denying data requests were required to provide BMJ with a brief explanation of their decision (Godlee and Groves 2012).
The movement from the ICMJE policy on trials registration to the more demanding policy of BMJ on patient-level data sharing reflects the ongoing evolution in the data-sharing environment. Journal policies fostering improved clinical trials data access have encouraged broader recognition of the need for open science and have moved the scientific community toward greater acceptance of data sharing. While such journal efforts undoubtedly have fostered the current climate of growing clinical research transparency, journal policies only reach studies whose results are intended for publication, thus indirectly favoring positive or noninferiority trial results at the expense of negative trial results. To take advantage of the full range of benefits offered by greater access to accumulated scientific knowledge, more inclusive data-sharing initiatives have been undertaken voluntarily by private industry, government, and nonprofit sponsors. Industry has played a prominent role in data sharing, in part as a response to media reports of industry’s resistance to open data practices. Specific industry sponsors faced significant public censure after studies published in 2013 revealed instances of selective reporting and “cherry-picking” of results; indeed, according to one report, only half of clinical trial results were published in full, and positive-result trials were published twice as often as negative-result trials (Goldacre 2013). Media attention intensified following reports that some pharmaceutical companies had planned to enlist patient advocates to fight EMA transparency efforts (Sample 2013). At the same time, some industry sponsors have willingly taken steps toward clinical trials data sharing. With industry responding to public pressure and changing data-sharing expectations, voluntary mechanisms to permit controlled access to clinical trials data have proliferated, moving data-sharing initiatives from regulation induced to voluntary.
These data-sharing initiatives present a wide diversity in model structures as well as the underlying terms of access. There are single- and multisponsor initiatives, academic–private partnerships, and public–private models built either around specific sponsors’ data or select conditions. The access requirements also vary, with some conducting a more rigorous evaluation of a requestor’s proposal than others, and some requiring executed data-use agreements. A diverse sample of noteworthy initiatives, selected for their varying approaches to data sharing, is described further in the following sections.
B. Company-Specific Platforms
One of the earliest company-specific efforts to increase transparency in clinical trials was undertaken by GlaxoSmithKline. In 2004, GSK became the first pharmaceutical company to introduce a publicly accessible clinical trial database, providing data from GSK’s clinical research studies (GSK 2013). In late 2012, GSK announced plans to make anonymized participant-level data available to qualified researchers through a controlled access model. To ensure that data is used for valid scientific purposes, the model requires interested researchers to submit a study protocol, certify the presence of a statistician on the research team, and commit to publish research results. Safeguarding against potential for bias in granting access to data, all decisions are made by an independent review panel consisting of experts appointed and compensated—but not controlled—by GSK. If a request is approved, the researchers gain access to raw data sets with specifications, updated study protocols, clinical study reports, annotated case report forms, reporting and analysis plans, and clinical study reports. Studies dating back to 2000 are made available if a drug is approved by regulators or has been terminated from development and the study has been accepted for publication (GSK 2014). In furthering its commitment to transparency, the GSK model informs requestors of the reasons for proposal acceptance/denial and makes public relevant statistics on the number of data requests submitted and actions taken.
GSK’s approach proved popular among other industry entities, and by July 2014 the GSK platform had evolved to become a multisponsor initiative hosting data from Bayer, Boehringer Ingelheim, GSK, Lilly, Novartis, Roche, Sanofi, and ViiV Healthcare, accessed at https://www.clinicalstudydatarequest.com (ClinicalStudyDataRequest.com 2014). While the sponsors utilize different guidelines for the inclusion of a study in the database and vary in anonymization standards, the type of shared information, review criteria, process, and data use agreements for enquiries are generally similar. One limitation of the system is that data must be analyzed on the system platform and cannot be downloaded; therefore, only data sets within the platform, and not external data sets, can be combined.
C. Disease-Specific Platforms
While GSK’s sharing model is an example of a platform built around a specific company, other initiatives have undertaken steps to organize and share data from clinical trials on specific diseases. For example, Parkinson’s Progression Marker Initiative (PPMI), an international partnership between mostly university-based clinical sites working toward identifying progression biomarkers in Parkinson’s disease, grants qualified researchers access to download clinical, biological, and imaging data from PPMI trials. Another disease-specific initiative gaining prominence is Project Data Sphere of the CEO Roundtable on Cancer, first announced in early 2013.
Unlike the GSK controlled-access initiative, Project Data Sphere can be characterized as a broad access model. Seeking to accelerate the pace of cancer research by facilitating broader sharing, Project Data Sphere attempts to create an independent online research repository of comparator arm data from both academic and industry-sponsored studies (Beetsch 2013). Data sets from various sponsors are integrated, including de-identified patient level data made accessible in a firewalled environment (Hugh-Jones 2013). Researchers granted access after meeting basic expertise requirements and agreeing to terms of use can search and filter studies based on such criteria as sponsor, tumor type, region, age group, study type, and cancer stage (Hugh-Jones 2013). Project Data Sphere does not utilize an applicant review panel and does not require a study proposal. It depends on voluntary data contributions from study sponsors, which are promised the benefits of precompetitive mutual data insights and the resulting improved productivity, cost savings, and increased citation rates (Hugh-Jones 2013). A significant limitation of Project Data Sphere is that it provides only nongenomic comparator arm data, at least in these initial stages. While Project Data Sphere generated significant support in its planning stages, it remains to be seen whether the initiative will enjoy sufficient popularity after its launch and gather the requisite critical mass of trials to make the database worthwhile.
D. Academic–Private Partnership Platforms
The Yale University Open Data Access (YODA) Project is an example of a successful data-sharing initiative centering on a partnership between academia and industry. YODA owes its start to a public controversy. In 2011, media reports disclosed that researchers with financial ties to the medical device company Medtronic had appeared to underreport adverse events and to exaggerate the efficacy results in clinical trials of Medtronic’s rhBMP-2 spinal fusion surgery product marketed as “Infuse” (Stanton 2013). In the wake of these reports, Medtronic engaged two groups of researchers at Yale to conduct an independent review of complete data sets from the Infuse clinical trials. Following completion of the initial review, Yale announced that, in its continuous partnership with Medtronic, YODA would make available clinical trials data from seventeen rhBMP-2 trials to qualified researchers who wished to advance scientific knowledge. A controlled-access model, YODA requires interested researchers to submit a study protocol, a certification or waiver of IRB approval, a conflict of interest statement, and a signed data use agreement (CORE 2014). Approved requesters receive data via a secured file transfer system and have one year to complete their study or apply for an extension.
Utilizing the experience gained from its cooperation with Medtronic, in January 2014, YODA announced a new initiative with Johnson & Johnson (J&J). Under the new partnership, a J&J subsidiary, Janssen Research and Development LLC, will release its clinical trials data, including de-identified patient-level data and clinical study reports, to qualified researchers (J&J 2014). While J&J will participate in establishing guidelines for reviewing requesters, all decision making on data access will be independently carried out by the university team (Peart 2014). YODA’s collaboration with J&J illustrates the potential of academic–industry partnerships.
E. Public Organization–Specific Platforms
A final category of platforms includes data-sharing projects initiated and run by public organizations. For example, the Immune Tolerance Networks’s (ITN) TrialShare portal, operated by an NIH-sponsored international clinical research consortium, provides the research community access to clinical trials data and biological specimens from ITN-funded studies, as well as auxiliary tools for data visualization to assist in interactive exploratory analyses (ITN 2013). The National Heart Blood and Lung Institute (NHBLI) BioLINCC project similarly shares biological specimen and participant-level data from NHBLI-sponsored trials with researchers who meet basic scientific qualifications. Lastly, the Laboratory of Neuro Imaging (LONI) project, partially supported by NIH, is a large-scale collaborative effort aiming to provide researchers with an interactive visualization environment for safely integrating, archiving, and querying neuro imaging data and associated patient data (Dinov et al. 2006; LONI).
A critical examination of participant-level data transparency efforts reveals that data sharing is possible and desirable and that the inherent risks can be mitigated. The move toward broader clinical trials data access reinforces the need for a sharing regime that promotes the potential benefits of transparency while including adequate safeguards against risks. It is noteworthy that, to date, major pharmaceutical companies have made the most visible and robust efforts to provide access to clinical trial data. Over time, small biotechnology companies, academicians, not-for-profit sponsors, and government agencies should develop mechanisms and secure resources to implement data-sharing initiatives. Funders of clinical trials (e.g., NIH, foundations) that are performed by academic investigators should anticipate and resource the incremental financial costs of data sharing, including development of data standards, metadata, and hosting the data. As the real power of data sharing involves combining different data sets, an early commitment to data standards and a common taxonomy are important for such interoperability. Finally, the participants themselves, patient groups, and patient advocates are becoming increasingly sophisticated and capable of planning and executing their own research (often termed “citizen science” or “participant-led research”); data availability will be important for these efforts as well. Well-tailored data-sharing mechanisms, with careful protections for participant privacy and private commercial interests, have the potential to revolutionize the pace of clinical research, facilitating the finding of new treatments and the promotion of public health.
NOTE
Alla Digilova and Rebecca Li are co–first authors.
REFERENCES
AbbVie, Inc. v. European Medicines Agency, Case T-44/13. Action brought on January 29, 2013 (March 1, 2013).
De Angelis, Catherine, Jeffrey M. Drazen, Frank A. Frizelle, Charlotte Haug, John Hoey, Richard Horton, et al. 2004. “Clinical Trial Registration: A Statement from the International Committee of Medical Journal Editors.” New England Journal of Medicine 351:1250–51.
Dinov, Ivo D., Daniel Valentino, Bae Cheol Shin, Fotios Konstantinidis, Guogang Hu, Allan MacKenzie-Graham, et al. 2006. “LONI visualization environment.” Journal of Digital Imaging 19(2):148–58.
European Medicines Agency (EMA). 2014. “Regulation (EU) No 536/2014 of the European Parliament and of the Council on Clinical Trials on medicinal products for human use, and repealing Directive 2001/20/EC.”
Godlee, Fiona. 2012. “Clinical Trial Data for All Drugs in Current Use.” British Medical Journal 345:e7304.
Godlee, Fiona and Trish Groves. 2012. “The New BMJ Policy on Sharing Data from Drug and Device Trials.” British Medical Journal 345:e7888.
Harrison, Charlotte. 2012. “GlaxoSmithKline Opens the Door on Clinical Data Sharing.” Nature Reviews Drug Discovery 11:891–92.
InterMune UK Ltd. v. European Medicines Agency, Case T-73/13. Action brought on February 11, 2013 (Apr. 5, 2013).
Kesselheim, A. S. and Mello, M. M. 2007. “Confidentiality Laws and Secrecy in Medical Research: Improving Public Access to Data on Drug Safety.” Health Affairs 26(2):483–91.
Mello, Michelle M., Jeffrey K. Francer, Marc Wilenzick, Patricia Teden, Barbara E. Bierer, and Mark Barnes. 2013. “Preparing for Responsible Sharing of Clinical Trial Data.” New England Journal of Medicine 369(17):1651–58.
U.S. Food and Drug Administration (FDA). 2013. “Availability of Masked and De-identified Non-Summary Safety and Efficacy Data; Request for Comments.” 78 Fed. Reg. 33, 421–23.
World Medical Association (WMA). 2013. “World Medical Association Declaration of Helsinki: Ethical Principles for Medical Research Involving Human Subjects.” Journal of the American Medical Association 310(20):2191–94.
