There is increasing interest to make primary data from published research publicly available. We aimed to assess the current status of making research data available in highly-cited journals across the scientific literature.
Methods and Results
We reviewed the first 10 original research papers of 2009 published in the 50 original research journals with the highest impact factor. For each journal we documented the policies related to public availability and sharing of data. Of the 50 journals, 44 (88%) had a statement in their instructions to authors related to public availability and sharing of data. However, there was wide variation in journal requirements, ranging from requiring the sharing of all primary data related to the research to just including a statement in the published manuscript that data can be available on request. Of the 500 assessed papers, 149 (30%) were not subject to any data availability policy. Of the remaining 351 papers that were covered by some data availability policy, 208 papers (59%) did not fully adhere to the data availability instructions of the journals they were published in, most commonly (73%) by not publicly depositing microarray data. The other 143 papers that adhered to the data availability instructions did so by publicly depositing only the specific data type as required, making a statement of willingness to share, or actually sharing all the primary data. Overall, only 47 papers (9%) deposited full primary raw data online. None of the 149 papers not subject to data availability policies made their full primary data publicly available.
A substantial proportion of original research papers published in high-impact journals are either not subject to any data availability policies, or do not adhere to the data availability instructions in their respective journals. This empiric evaluation highlights opportunities for improvement.
Citation: Alsheikh-Ali AA, Qureshi W, Al-Mallah MH, Ioannidis JPA (2011) Public Availability of Published Research Data in High-Impact Journals. PLoS ONE 6(9): e24357. https://doi.org/10.1371/journal.pone.0024357
Editor: Isabelle Boutron, University Paris Descartes, France
Received: May 17, 2011; Accepted: August 6, 2011; Published: September 7, 2011
Copyright: © 2011 Alsheikh-Ali et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: These authors have no support or funding to report.
Competing interests: The authors have declared that no competing interests exist.
The observations of scientists as coded in their primary data constitute a central commodity in the scientific enterprise . Reproduction of research findings and further exploration of related hypotheses require access to these primary data, and their public availability has been a concern for all stakeholders of the scientific process, including regulatory and funding agencies, journal editors, individual researchers, and patients , , , , , . Recently, efforts have converged to encourage making data, protocols, and analytical codes available, as part of the growing movement of reproducible research , , . The benefits and challenges of public data availability and data sharing have long been hotly discussed in the scientific community , , . Recent analyses have empirically highlighted deficiencies in the practice of making primary data and protocols available in peer-reviewed publications, , . These analyses, however, have focused on either a particular discipline or area of research or were limited to a single journal. To date, there has not been an empiric evaluation of public availability of primary data and related material and protocols across diverse scientific fields or journals. We aimed to assess the current status of these practices in the most highly-cited journals across the scientific literature.
We examined the 50 journals with the highest impact factor according to the Journal Citation Reports (Science edition 2007) issued in the Thompson-Institute for Scientific Information Web of Knowledge. Journals that exclusively publish review articles were not included. For each journal, we also reviewed the first 10 original research papers published in 2009.
For each journal we documented the policies related to public availability and sharing of data, where available, and as stated in the instructions to authors on the journal's website (accessed May 2009). This was done by one investigator (AA) and verified by a second (MA). Each paper was reviewed by going through the text, supplementary material and the links available on the online version. We recorded information on country of first and corresponding authors, funding sources, data links and accession numbers, and whether the paper was based on data covered by a journal policy (e.g. paper with microarray data published in a journal requiring public deposition of microarray data). Data extraction from the 500 papers was done by a single investigator (WQ) with cross-checking in the first 50 papers by two investigators (AA and MA). This information was collected in the months of July and August of 2010. Online information was considered missing if links were not available when checking on 2 separate occasions 2 weeks apart. We compared the impact factor of journals where provision of materials and protocols was a condition of publication versus journals with non-binding instructions or no instructions at all using the Kruskall-Wallis analysis of variance test. We compared the proportion of papers depositing full primary data online by status of US government funding and by geographic origin of corresponding author (US versus non-US) using the Chi square test.
Of the 50 highest impact factor journals publishing original research, 44 (88%) had a statement in their instructions to authors related to public availability and sharing of data from submitted manuscripts (Appendix S1). However, there was wide variation in journal requirements, ranging from requiring the sharing of all primary data related to the research to just including a statement in the published manuscript that data can be available on request. Some specific types of data had very high frequency of requirement for public deposition. This included public deposition of primary microarray, nucleic acid and protein sequencing data, and macromolecular structures which was required in 36/50 (72%), 40/50 (80%), 39/50 (78%), and 29/50 (58%) journals (Figure 1). Materials and protocols used in published experiments were required to be made available upon request to qualified researchers by 33/50 (66%) and 23/50 (46%) journals, respectively.
When instructing authors on data or material sharing, journals used different phrases to indicate how strict these requirements were. Less than half (22/50, 44%) of the journals explicitly indicated that making materials and/or protocols of the published findings (regardless of method and technology employed) available to other qualified investigators was a condition of publication. In their instructions to authors regarding data sharing, these journals used language such as “with minimal restriction . . . in a timely manner” (e.g. Cell family of journals), “non-compliance … may result in denial of future rights to publish” (e.g. Plant Cell), or “…a condition of publication . . . is . . to make materials, data and associated protocols promptly available . . . without preconditions” (e.g. Nature family of journals). An additional 44% (22/50) of the journals had either a non-binding statement encouraging authors to make their data available (e.g. using language like “it is the responsibility of the authors” or “authors are encouraged to . . .”), or required a data-sharing statement by the authors indicating willingness to share (Annals of Internal Medicine and British Medical Journal). The remainder 6 journals (12%) had no specific instructions to authors related to data availability. Journals where provision of materials and protocols was a condition of publication had higher impact factors compared to journals with non-binding instructions or no instructions at all (median [25th, 75th percentiles]: 15.14 [11.09, 19.78] versus 12.68 [9.72, 15.98] versus 9.83 [9.13, 11.05], P = 0.04 by Kruskal-Wallis analysis of variance).
Of the 500 papers we reviewed (Appendix S2), 149/500 (30%) were not subject to any data availability policy (60 were published in journals without a specific data sharing statement and 89 additional papers contained data not covered by the specific public deposition policy in their journals). Of the remaining 351 papers that were covered by some data availability policy (with or without data deposition as a condition of publication), 208/351 papers (59%) did not fully adhere to the data availability instructions of the journals they were published in, most commonly (73%) by not publicly depositing microarray data. The other 143 papers (of the 351 covered by some data availability policy) that adhered to the data availability instructions of the journals did so by publicly depositing only the specific data type as required, making a statement of willingness to share, or actually sharing all the primary data. Overall, only 47/500 papers (9%) did deposit full primary raw data online; and we were able to verify access in all 47 (sites were accessed in July and August 2010). None of the 149 papers not subject to data availability policies made their full primary data publicly available.
Among papers covered by some data availability policy (n = 351), the proportion depositing full primary data online was not different when US government funding (e.g. National Institutes of Health or National Science Foundation) was listed (15% versus 12% when US funding was not listed, P = 0.42) or the corresponding author was from a US institution (16% versus 11% for non-US corresponding authors, P = 0.20).
The present overview of highly cited journals highlights three main features of the current status of data availability practices in this high impact scientific literature. First, there are heterogeneous instructions to investigators publishing in high impact journals, with some journals requiring public data availability as a condition for publication, others encouraging data sharing but having no binding instructions, and a few journals having no specific instructions at all. Second, nearly a third of the examined sample of 500 papers were not subject to any data availability policies, either because they were published in journals without such policies or with specific policies that do not cover the primary data upon which the research was based. Third, even when research is published in journals with specific instructions regarding data availability, more than half of publications did not adhere to the data availability instructions in their respective journals.
Our findings present a snapshot of data availability practices in recent literature. While the papers we reviewed were from 2009, it is unlikely that the situation has changed much over the past year, and we therefore believe that the present findings represent reasonably well the current state of the literature. Moreover, since the papers we reviewed were likely submitted 6-12 months prior to our recording of journal policies, some journals may have adopted data sharing policies in the interim, hence inflating our estimate of lack of adherence to data sharing policies. However, it is doubtful that policies related to data sharing have changed substantially over such a short period of time. We also focused our analysis on high impact journals, since the research that they publish has a pivotal role in the evolution of scientific investigation and it is essential that this pivotal research is reproducible. It is not likely that data availability practices are more common and more efficient in other journals with lower impact factor - the opposite seems more plausible, if anything. Therefore the present findings may well overestimate the prevalence of effective data sharing among investigators publishing across all peer-reviewed journals. In fact, some types of biomedical studies, in particular traditional epidemiological/observational investigations, may be underrepresented in our sample as compared with molecular and other clinical research. Some of these types of underrepresented studies have no established history of public data repositories and thus primary data availability may be a more critical deficiency in these fields. It is also worth noting that the association between higher impact factor and conditioning publication upon provision of materials/protocols may be confounded by type of journal, as experimental/basic science journals that typically have such conditions tend to have higher impact factors.
While this analysis highlights an important element of data sharing, that of public availability of primary data, there are other elements not evaluated here but still important to make the data sharing culture functional and efficient. For example, a statement of willingness to share raw data by the primary investigators does not always translate into true availability of data when requested by independent scientists . Empirical studies suggest that data withholding is not uncommon in the scientific community and may be influenced by industry relationships, perceptions of proprietary information and scientific priority, lack of resources, and personal investigator training and stances towards data sharing, , . Moreover, while all data web links of full primary datasets were verified as functioning in our analysis, this may reflect the temporal proximity of our analysis to the publication date of the articles, and some of these links may become unavailable a few years later .
Legislation to make results of clinical trials publicly available within one year of study completion may promote the culture of transparency in clinical trials research, but at present such legislation does not mandate making raw data from clinical trials publicly available . Indeed, widespread availability of clinical trial data may be hampered by financial incentives of journals to publish industry-sponsored trials, many of which may be bound by confidentiality agreements , , . Data sharing may be enhanced when granting agencies require investigators to share data but regulatory barriers remain , .
Finally, for data that was made available by investigators, we did not attempt to replicate their findings. Even when data are publicly available, published results are often not reproducible by independent investigators due to incomplete annotation or specification of data processing and analyses .
This empiric evaluation highlights opportunities for improvement. Journals should adopt more routinely policies for data sharing, expanding the types of data that are subject to public sharing policies with the ultimate target of covering all types of data. Moreover, it is essential to develop mechanisms for journals to ensure that existing data availability policies are consistently followed by researchers and published research findings are easily reproducible.
Summary of journal policies as they relate to data sharing.
Conceived and designed the experiments: JI AA-A. Performed the experiments: AA-A WQ MA-M JI. Analyzed the data: AA-A WQ. Wrote the paper: AA-A WQ MA-M JI.
- 1. Young NS, Ioannidis JP, Al-Ubaydli O (2008) Why current publication practices may distort science. PLoS Med 5: e201.NS YoungJP IoannidisO. Al-Ubaydli2008Why current publication practices may distort science.PLoS Med5e201
- 2. Guttmacher AE, Nabel EG, Collins FS (2009) Why data-sharing policies matter. Proc Natl Acad Sci U S A 106: 16894.AE GuttmacherEG NabelFS Collins2009Why data-sharing policies matter.Proc Natl Acad Sci U S A10616894
- 3. (2009) Data's shameful neglect. Nature 461: 145.2009Data's shameful neglect.Nature461145
- 4. Groves T (2010) The wider concept of data sharing: view from the BMJ. Biostatistics 11: 391–392.T. Groves2010The wider concept of data sharing: view from the BMJ.Biostatistics11391392
- 5. Sommer J (2010) The delay in sharing research data is costing lives. Nat Med 16: 744.J. Sommer2010The delay in sharing research data is costing lives.Nat Med16744
- 6. Godlee F (2009) We want raw data, now. BMJ 339: b5405.F. Godlee2009We want raw data, now.BMJ339b5405
- 7. Hrynaszkiewicz I, Altman DG (2009) Towards agreement on best practice for publishing raw clinical trial data. Trials 10: 17.I. HrynaszkiewiczDG Altman2009Towards agreement on best practice for publishing raw clinical trial data.Trials1017
- 8. Baggerly K (2010) Disclose all data in publications. Nature 467: 401.K. Baggerly2010Disclose all data in publications.Nature467401
- 9. Hrynaszkiewicz I, Norton ML, Vickers AJ, Altman DG (2010) Preparing raw clinical data for publication: guidance for journal editors, authors, and peer reviewers. BMJ 340: c181.I. HrynaszkiewiczML NortonAJ VickersDG Altman2010Preparing raw clinical data for publication: guidance for journal editors, authors, and peer reviewers.BMJ340c181
- 10. Vickers AJ (2006) Whose data set is it anyway? Sharing raw data from randomized trials. Trials 7: 15.AJ Vickers2006Whose data set is it anyway? Sharing raw data from randomized trials.Trials715
- 11. (2010) Sharing public health data: necessary and now. Lancet 375: 1940.2010Sharing public health data: necessary and now.Lancet3751940
- 12. Samet JM (2009) Data: to share or not to share? Epidemiology 20: 172–174.JM Samet2009Data: to share or not to share?Epidemiology20172174
- 13. Pisani E, AbouZahr C (2010) Sharing health data: good intentions are not enough. Bull World Health Organ 88: 462–466.E. PisaniC. AbouZahr2010Sharing health data: good intentions are not enough.Bull World Health Organ88462466
- 14. Ioannidis JP, Allison DB, Ball CA, Coulibaly I, Cui X, et al. (2009) Repeatability of published microarray gene expression analyses. Nat Genet 41: 149–155.JP IoannidisDB AllisonCA BallI. CoulibalyX. Cui2009Repeatability of published microarray gene expression analyses.Nat Genet41149155
- 15. Savage CJ, Vickers AJ (2009) Empirical study of data sharing by authors publishing in PLoS journals. PLoS One 4: e7078.CJ SavageAJ Vickers2009Empirical study of data sharing by authors publishing in PLoS journals.PLoS One4e7078
- 16. Wicherts JM, Borsboom D, Kats J, Molenaar D (2006) The poor availability of psychological research data for reanalysis. Am Psychol 61: 726–728.JM WichertsD. BorsboomJ. KatsD. Molenaar2006The poor availability of psychological research data for reanalysis.Am Psychol61726728
- 17. Blumenthal D, Campbell EG, Anderson MS, Causino N, Louis KS (1997) Withholding research results in academic life science. Evidence from a national survey of faculty. JAMA 277: 1224–1228.D. BlumenthalEG CampbellMS AndersonN. CausinoKS Louis1997Withholding research results in academic life science. Evidence from a national survey of faculty.JAMA27712241228
- 18. Campbell EG, Clarridge BR, Gokhale M, Birenbaum L, Hilgartner S, et al. (2002) Data withholding in academic genetics: evidence from a national survey. Jama 287: 473–480.EG CampbellBR ClarridgeM. GokhaleL. BirenbaumS. Hilgartner2002Data withholding in academic genetics: evidence from a national survey.Jama287473480
- 19. Blumenthal D, Campbell EG, Gokhale M, Yucel R, Clarridge B, et al. (2006) Data withholding in genetics and the other life sciences: prevalences and predictors. Acad Med 81: 137–145.D. BlumenthalEG CampbellM. GokhaleR. YucelB. Clarridge2006Data withholding in genetics and the other life sciences: prevalences and predictors.Acad Med81137145
- 20. Evangelou E, Trikalinos TA, Ioannidis JP (2005) Unavailability of online supplementary scientific information from articles published in major journals. FASEB J 19: 1943–1944.E. EvangelouTA TrikalinosJP Ioannidis2005Unavailability of online supplementary scientific information from articles published in major journals.FASEB J1919431944
- 21. Bian Z, Wu T (2010) Legislation for trial registration and data transparency. Trials 11: 64.Z. BianT. Wu2010Legislation for trial registration and data transparency.Trials1164
- 22. Smith R (2006) Conflicts of interest: how money clouds objectivity. J R Soc Med 99: 292–297.R. Smith2006Conflicts of interest: how money clouds objectivity.J R Soc Med99292297
- 23. Lundh A, Barbateskovic M, Hrobjartsson A, Gotzsche PC (2010) Conflicts of interest at medical journals: the influence of industry-supported randomised trials on journal impact factors and revenue - cohort study. PLoS Med 7: e1000354.A. LundhM. BarbateskovicA. HrobjartssonPC Gotzsche2010Conflicts of interest at medical journals: the influence of industry-supported randomised trials on journal impact factors and revenue - cohort study.PLoS Med7e1000354
- 24. Gotzsche PC, Hrobjartsson A, Johansen HK, Haahr MT, Altman DG, et al. (2006) Constraints on publication rights in industry-initiated clinical trials. JAMA 295: 1645–1646.PC GotzscheA. HrobjartssonHK JohansenMT HaahrDG Altman2006Constraints on publication rights in industry-initiated clinical trials.JAMA29516451646
- 25. Miller JD (2010) Sharing clinical research data in the United States under the Health Insurance Portability and Accountability Act and the Privacy Rule. Trials 11: 112.JD Miller2010Sharing clinical research data in the United States under the Health Insurance Portability and Accountability Act and the Privacy Rule.Trials11112
- 26. Vickers AJ (2011) Making raw data more widely available. BMJ 342: d2323.AJ Vickers2011Making raw data more widely available.BMJ342d2323