Note: This blog was written by Renee Hoch, Senior Editor, Team Manager for PLOS ONE’s Publication Ethics Editorial team
The integrity of image data in primary research articles has emerged as a major concern in publication ethics. Studies querying the frequency of image concerns have indicated that 4-25% of published research articles in the biomedical and life sciences may include unreported image splicing, duplications, and/or manipulations with potential repercussions for the reliability of the results [1-4]. These issues can stem from honest errors, efforts to beautify or clarify the results, poor data management, poor knowledge as to best practice standards in image data reporting, or fraudulent misrepresentation of experimental results. Whatever the root cause, image integrity issues can damage the reliability of – and public trust in – the scientific literature.
What can journals do?
Up to now the majority of image concern cases at PLOS ONE and PLOS Biology have been raised post-publication, although some are spotted during peer review. We follow up on such issues regardless of when they are raised, but it would benefit authors, journals, and readers to try to identify and address more of these concerns prior to publication. To this end, as discussed recently at the 6th World Conference on Research Integrity, some journals have implemented dedicated pre-publication image screening; developers are working on software tools to help identify issues such as splicing and duplications; and a group at the Headt Center in Berlin is developing an Image Integrity Database  that can be used to test new software tools. Solutions are needed which can be applied at scale within the context of journal checks and/or peer review.
At PLOS ONE and PLOS Biology, we are taking a policy-based approach to proactively address image integrity for blot and gel results: we now require that authors submit the underlying raw images to support all blot and gel image data reported in a submission’s figures or supporting information files. The raw data will be assessed during peer review (PLOS ONE) or before acceptance (PLOS Biology), and if a manuscript is accepted the underlying image data will be published as supporting information or provided via a public data repository.
Why require raw blot and gel data?
Our new policy focuses on blot and gel reporting because these types of images comprise the most prevalent category of image concerns that our editorial teams handle, and the raw image data underlying these figures are particularly informative in clarifying the reliability of the reported results. Under the new policy, editors and reviewers will have access to the underlying image data before publication so that they can verify the reported results and consider these data when making decisions as to whether the work complies with the journals’ guidelines and publication requirements. With the raw data, editors and reviewers will be better equipped to identify any concerns about these results in the context of peer review, so that authors can address questions about their data and resolve honest data reporting errors prior to publication. With this new policy and the support of our authors, Academic Editors, and reviewers, we aim to strengthen the integrity and reproducibility of our published articles.
Figure 1. Availability of image data requested in post-publication image concern cases
(based on PLOS ONE cases resolved 1/2017-5/2019)
The new policy will bring our data reporting requirements for blot and gel results into better alignment with PLOS’ and the greater community’s open data-sharing objectives. Since 2014, PLOS’ Data Availability Policy has required that authors provide the underlying data supporting an article’s results. For image data, our policy has specified that authors provide “sample” images to support the results, but has not provided clear instructions about our expectations for sample images. One could consider published figure panels to be representative sample images but these rarely include the original image data: images are typically cropped, adjusted, and in some cases manipulated (stretched, spliced, contrast adjusted, etc.) during figure preparation. While such alterations may be acceptable if transparently reported and abiding by best practice standards, they can also lead to questions about a figure’s integrity, make it difficult to assess the results, and in some cases lead to loss of important information such as about reagent specificity or experimental controls. The original data play an essential role in clarifying such issues when raised, but unfortunately authors are often unable to provide these data after publication (Figure 1 and ). Our new policy clarifies our blot/gel image data requirements in detail, and aims to ensure the availability of blot/gel image data after an article’s publication. We have also updated our blot/gel reporting and figure preparation guidelines to increase awareness of best practices in blot and gel presentation and support transparency in data reporting.
We sincerely thank all who provided advice on these recent PLOS ONE and PLOS Biology updates, and we thank our authors in advance for working with us towards these higher image data reporting standards. We appreciate that the new policy may require extra work, but we hope that these efforts will positively impact the integrity and reproducibility of the published record and ultimately benefit the scientific community.
- Bik EM, Casadevall A, Fang FC. The prevalence of inappropriate image duplication in biomedical research publications. mBio 7(3):e00809-16. (2016) doi:10.1128/mBio.00809-16. http://mbio.asm.org/content/7/3/e00809-16.full
- Oksvold, MP. Incidence of Data Duplications in a Randomly Selected Pool of Life Science Publications. Science and Engineering Ethics 22(2): 487-496. (2015) https://link.springer.com/article/10.1007%2Fs11948-015-9668-7
- Bik EM, Fang FC, Kullas AL, Davis RJ, Casadevall A. Analysis and Correction of Inappropriate Image Duplication: the Molecular and Cellular Biology Experience. Mol Cell Biol. 38(20). pii: e00309-18. (2018) doi: 10.1128/MCB.00309-18. https://www.ncbi.nlm.nih.gov/pubmed/30037982
- Christopher J. Systematic fabrication of scientific images revealed. FEBS Letters 592(18): 3027-3029. (2018) https://febs.onlinelibrary.wiley.com/doi/10.1002/1873-3468.13201
- Beck T. Combating image misuse in science: new Humboldt database provides “missing link”. Elsevier Connect Oct. 9, 2018. https://www.elsevier.com/connect/combating-image-misuse-in-science-new-humboldt-database-provides-missing-link
- Vines TH, Albert AYK, Andrew RL, Débarre F, et al. The Availability of Research Data Declines Rapidly with Article Age. Current Biology 24(1): 94-97. (2014) https://www.sciencedirect.com/science/article/pii/S0960982213014000