Dr Elisabeth Bik is the founder of Microbiome Digest, an almost daily compilation of scientific papers in the rapidly growing microbiome field, and in March 2019, she became a microbiome and science integrity consultant. She can often be found discussing science papers on Twitter as @MicrobiomDigest, writing for her blog ScienceIntegrityDigest or searching the biomedical literature for inappropriately duplicated or manipulated photographic images and plagiarized text. The Biochemist spoke to Elisabeth as part of our issue on research culture.

I like gardening – not the fruit or vegetable growing kind, because the squirrels always get there first. But I love pruning shrubs and repairing sprinkler systems. I also go twice a week to Bombay Jam, a cardio-fitness class on Bollywood dance music, and I swim laps once a week. But most of the time, I sit behind my computer working on science misconduct. I do that about 12 h a day.

In elementary school, I enjoyed learning about animals and plants, and I decided I wanted to study biology. I never changed my mind. In college, I especially enjoyed molecular biology, which was a very new field at that time, and microbiology, an old field, but that got a huge boost because of new molecular techniques. It was fascinating to me to be able to study microscopically small organisms without needing to culture them.

Science for me is always learning new things, designing good studies, and also being critical. If people make fantastic claims without citations or proof, it is good to be sceptical and ask critical questions.

I heard about scientific misconduct around 2013 and was immediately fascinated about this topic. I started by searching for plagiarized sentences and paragraphs by putting sentences between quotes in Google Scholar and immediately found several examples. Later, by accident, I discovered a photo of a Western blot that was used twice within the same paper, but to represent two different experiments. I then switched to searching for those image duplications by just scanning biomedical papers. This was my free-time project for many years, in the evenings or in the weekend while I was working at Stanford University as a scientist, and later while I worked in industry. In March 2019, I decided to quit my job and do this work full time.

One of the reasons is that there is more and more pressure on scientists to publish. Scientific success is increasingly measured by the amount of scientific papers that a person produces per year. In reality, doing research can be very slow and painful, with lots of failures and wrong turns. If a grad student, post-doc or assistant professor is told they need to produce a paper that month or year, the temptation to falsify or fabricate results can be very high, especially if that means keeping your job. In recent years, medical schools and hospitals in some countries have had unrealistic demands for students or doctors to publish at least one paper in order to graduate or get a promotion, while not allowing these people time to actually do research. That might have led to a recent increase in fabricated papers from particular hospitals. In most of these cases, the people are only an author on a single biomedical paper; they do not become a researcher, but just needed that one paper for their career.

We were not able to confirm that in our analysis of 346 individual papers with image problems (study published here: https://link.springer.com/article/10.1007/s11948-018-0023-7). However, this was a slightly different set of papers, namely papers by different persons with hardly any shared authors.

The US Office of Research Integrity (ORI) cases mentioned in the 2013 study linked above are mainly from clusters of cases, where multiple papers by the same person were found to have problems. In addition, the Retraction Watch leaderboard, which lists the people with the most retractions, contains mainly men as well: https://retractionwatch.com/the-retraction-watch-leaderboard/

So, maybe it is more correct to say that misconduct appears to happen by both genders, but that the serial fraudsters are mostly male.

Journals need to have strict guidelines about figure preparation, in particular about how to prepare gels and blots, and on what types of photo manipulation are allowed. Examples for those guidelines are not allowing splicing, or cropping Western blots too close to the bands, as well as to require submitting uncropped blots as supplementary data and explicitly prohibiting duplication of parts of photos. The Journal of Cell Biology (JCB) has long been a leader in requirements for figure preparation. They realized more than 15 years ago that stricter image rules needed to be part of their Instructions for Authors. This paper by Rossner and Yamada has been seminal in that respect: https://rupress.org/jcb/article/166/1/11/34064/What-s-in-a-picture-The-temptation-of-image

In my scan of 40 journals, JCB was by far the journal with the least amount of image problems. In the past couple of years, other journals are finally starting to follow JCB's lead.

In addition, manuscripts need to be better screened for image problems by specialized staff. Peer reviewers cannot be expected to find all problems, but it might help to add a question on the review checklist asking about suspicions in images. Many of these image problems can be found by investing five extra minutes in reviewing a paper.

I asked COPE to intervene in this particular case, but they decided that the journal took the right decision, since the manipulation was minor in their opinion. That is frustrating, but unfortunately very typical. Of the 800 or so cases I reported about 5 years ago, two thirds have not been touched at all. Only one third have been corrected, and a small subset has been retracted. The COPE guidelines are very vague in my opinion. As I proposed in a recent blog post (https://scienceintegritydigest.com/2020/01/08/oops-i-did-it-again/), there should be clear guidelines in terms of numbers of duplications allowed to be addressed by a correction vs retraction and much stricter rules about what counts as manipulation. The vagueness of the COPE guidelines, and the lack of journals' adherence to them, is contributing to cases of misconduct being continuously swept under the rug.

Yes, for sure. It is much easier now to change the results behind the computer, than it was 20, 30 years ago, when researchers had to bring their wet blots and gels to a professional photographer.

There is always a story of pressure to publish behind these cases, and most of these stories are sad. It can be a grad student who is being told that they need to have a result at a particular deadline or otherwise they will be kicked out of the programme. Or it is a postdoc on a temporary visa working in a famous research group who needs to get that publication or they will have to go back home without any. Or it is the assistant professor who had one successful paper in a high-tier journal and now has to live up to high expectations, but their research is not going very well. Or, as I mentioned above, it is a medical student who is required to write one research paper in order to get their degree, even though they are not interested in doing research and even if they have to work 80 hour per week taking care of patients. As a scientific community, success is obviously measured in terms of number of papers, but maybe we need to think about better ways to do this. Maybe we should not put as much pressure on the impact factor or numbers of peer-reviewed papers, but focus more on, e.g., preprints, including those with negative results.

I have heard several contradicting news stories, so I am not sure which ones are true. It would be great if the rule to publish at least one paper during medical school for students who want to focus on clinical work, not research, is scrapped.

Several cases come to mind, but I will only share some that have been retracted.

  1. A recent one that got retracted was a case where a patient was supposedly treated with a laser to remove some facial lesions. However, the before and 6 months after photos were suspiciously identical, except for the presence of the lesions. The authors admitted to manipulating it, and the paper is now retracted. Here it is: https://www.sciencedirect.com/science/article/pii/S0901502718303412 – see PubPeer: https://pubpeer.com/publications/7611C2EEE5789028895D8BD02A151E

  2. In another example, several Western blots contained duplicated bands and the institute concluded that these had been fabricated: https://pubpeer.com/publications/9D8E6F914E2B0CF873A82446CEEB0F and dx.doi.org/10.1371/journal.pone.0051309

  3. In a third example, several microscopy panels contained digitally edited and duplicated cells: https://pubpeer.com/publications/6D554774E0512108853685918D1AA1 and https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0055298

  4. Finally, this example also was retracted because of manipulated microscopy images. One of the best in my collection: https://pubpeer.com/publications/E471BDD2C63461C50C889707D407E8 and https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0164064

Several other people are doing this work too, although most people do it part-time and often using pseudonyms. But there are so many cases still undiscovered, not just recent ones, but also from 2000 and up, that we could use some more help for sure. If people want to participate, it helps to read what others are posting onto PubPeer.com for a while, to get a feeling for which types of problems can be reported and how to objectively word the allegations. It is best to stick to wording such as ‘image A looks unexpectedly similar to image B’, instead of stating that two images are identical or that fraud has been committed. For most people, it is safest to post under a pseudonym. Posting on PubPeer has the advantage of immediately flagging a paper for concerns, in particular for others who have the PubPeer browser extension installed. The more official route to report image concerns is to write to the journal (in single cases) or to the institute (if there are multiple cases from the same group), but often such allegations are ignored or take years to resolve.

2020 © Biochemical Society
2020