Analysis of Retractions from Hospitals in Mainland China from 2017-2019

Objective : The aim of this study is to analyze the features of retractions from hospitals in mainland China, and to discuss the causes of research misconduct by Chinese doctors. Research Methods : PubMed, Web of Science, and Retraction Watch Database were searched to collect eligible records and to extract characteristics of the included entries, including publishers and Open Access status of the journals involved, ORCID, PubPeer comments before the retraction, whether there are authors from Grade A, Third-class hospitals, and whether there are response or requirements from authors. Results : 521 retractions were included. Retractions were found primarily from authors of grade A, third-class hospitals, a limited regions, and published in journals with medium and high impact factor. The main reasons for retractions were Data Manipulation/Fabrication/Fraud(27.1%), Error by Author(19.9%), Plagiarism(16.7%), Self-Plagiarism(9.1%), Fake Peer Review(7.6%) and Forged Authorship(6.3%). Most of the retracted publications have neither ORCID nor PubPeer comments before their retraction. Conclusion : This is the first report focus on the retractions from hospitals in mainland China. The large number of retractions from Chinese hospitals in recent years is worrying. The results suggests that some retractions are related to third-parties. Some features of retractions are centralized, and it is difficult to evaluate the role of ORCID and PubPeer in the retractions during this period.


Introduction
In 2017, Tumor Biology, a journal once indexed by SCIE, retracted 107 papers from mainland China in a single action (1). In following months, the Ministry of Science and Technology of the People's Republic of China investigated 521 authors and, found 486 of them responsible for their wrongdoing leading to the massive retractions(2). This incident raised widespread concerns about the quality of biomedical research articles from mainland China and marked the beginning of a series of investigations from publishers and scientific community. Retraction of publications is a troublesome issues, especially for biomedical journals, which may bring many negative consequences for authors, institutions, and journals, such as affecting the credibility of academic journals and causing lack of trust in medical science (3). may improve the fairness of peer reviews before and after publication, but few studies have focused on Chinese authors who are from hospitals. It is not a reasonable assumption that researchers who graduated from the same university and work in the same institution, just because they have entered different research fields, will differ in research misconduct, or there are differences in the integrity of researchers entering different fields. Therefore, we should pay attention to the role of authors and institutions in research misconduct and retractions, and especially analyze the characteristics of some specific groups among them. In this study, we reviewed retractions that occurred in hospitals in mainland China from 2017 to 2019, and focus on features that may explain the reasons for research misconduct among Chinese doctors.

Material And Methods
We collected and aggregate data from three databases, PubMed, Web of Science, and Retraction database(Version: 1.0.5.5) provided by Retraction Watch. The search in PubMed is. The search in Retraction database is the affiliation is "hospital AND China" and the timespan for retraction is from January 1, 2017 to December 31, 2019. Two reviewers independently filtered the search results to remove duplicate and irrelevant results.
Two reviewers independently classified the reasons for each article being retracted and if there is response or requirements from author(s), and the conflict between the two reviewers was determined by the supervisor. We categorize the reasons for retractions based on Retraction Watch Database User Guide (8). Specifically, the reasons can be divided into: (1)Self-Plagiarism: including duplication of article, data, image, and text. (2)Plagiarism: The publisher made a clear allegation of plagiarism. China's official website (9,10). For a very small number of retracted articles and retractions that are not available from the publisher 's website, we try to obtain the historical version through SCI-HUB.
Statistical analysis was completed in R 3.6.2.

Results
We obtained 1,979 records from PubMed, 2,006 records from Web of Science Core Collection, 547 records from Retraction database, and screened 521 articles for inclusion in the study. We describe the features of retracted articles in Table 1. Retraction styles vary, and most retractions give clear but limited information. ideas in different periods, depending on whether they can win in this competition. But even those at a disadvantage in this race may not be willing to accept another set of qualitative and vague rules. As a result, doctors invest a lot of effort in writing articles, even if they will occasionally like the post that calls for cancellation of quantitative evaluation.
A common view is that academic journals should explain the reasons for retractions in more detail and in a more standardized way (15,16). In this study, sometimes we need some help from PubPeer comments to figure out the reason for retractions, because the retraction notice is not clear.
However, journals need the cooperation of authors to obtain sufficient evidence. In the case of insufficient evidence, journals often choose vaguely word retractions. In an intriguing example, the journal has announced that the experimental methods described by the authors were not able to get the results they reported, but they have not made up their minds to retract that article, and just advised readers to "interpret these results with caution" (17). Some retractions indicate that the author refused to respond to publisher inquiries, nevertheless, we observed that in 37 cases the editors still accused and retracted the articles. We suggest that hospitals should list refusals to respond to editor's inquiries and failure to find original experimental results as reasons for punishment, not just retractions.
Although the reasons for many retractions are not clearly given, we can still speculate on some reasons based on our research. Some authors seem unfamiliar with academic norms and do not have any sense of academic integrity. For example, a group of authors published four Meta-analyses that were highly similar except for the substance in 2014, and two of them were identified as selfplagiarized (18)(19)(20). The details mentioned in another retraction may explain this phenomenon. The authors clarified that they referred to templates to write their article, which from a training course on how to write Meta-analyses (21). Some retractions from hospitals in mainland China due to selfplagiarism, plagiarism, fake peer review, forged authorship and data manipulation/fabrication/fraud is confirmed to involve third parties, which has been mentioned by several retractions (22)(23)(24). In other words, we guess that many Chinese doctors are not able to complete some fraud independently.
These third parties usually charge high fees, and may use aggressive violations to help clients publish articles. In a case of forged authors and plagiarism, a third party got an unpublished manuscript from an agency for language editing, and used that to write an article for their clients, which brought tricky trouble for the real authors (25). In another case, it may be due to a lack of communication that a third party published the manuscript in another journal after the author had already published it (26). In other cases, the results were purchased by the author from a third party, and they can not ensure the reliability of their articles (27). For example, the editor stated in a retraction that the authors' contributions to the work were "unclear" in their perspective (28). A third party provided the same results to two research teams, and the articles of both teams solved the same research questions and reported the same data, results, and conclusions (29). Recent investigations have suggested the presence of paper mills (30). They use manuscript templates to produce a lot of similar articles, regular and similar results pictures, and submit these articles to various journals for his client. Some Chinese doctors may be involved, but this is a new discovery that received the attention of the public only in 2020 (31), and it is not yet reflected in the results we have collected in these three years. It is worth noting that there are tens of retractions due to ethical issues occurred, which mentioned similar reasons, and similarly worrying matters behind (32).
We counted several external features of retracted articles and retractions, and found that the retractions were concentrated in grade A, third-class hospitals, a limited regions, and medium and high-impact journals. Explanations of these phenomena require qualitative interviews to be published.
There are not many low impact factor articles, probably because the benefits are too small compared to the risk of research misconduct. The post-publish peer review website PubPeer provides an effective platform for senior scientists to point out that they have found research misconduct.
However, it is difficult to evaluate the role of PubPeer in these retractions from 2017 to 2019, as most articles(92.7%) do not have any comments posted before they were retracted. Only a small number of comments were answered by the author, and the author's response was generally decent. We found ORCID information in a few articles(6.5%), but most did not show any public information. Although we did not see the current role of ORCID in holding researchers responsible for their retractions, we can still be optimistic. We suggest ORCID to consider that the retraction records are showed and visible to the trusted parties or everyone by default on ORCID homepage, which can accurately increase the adverse consequences of research misconduct. As more and more researchers use ORCIDs, publishers can force authors to submit their ORCIDs and associated retraction records. Even if the hospital subscribes to a small academic database, it can review applicants' ORCID accounts to check if they have retracted manuscripts when hiring and promoting doctors. Hospitals should record the email addresses and ORCIDs of senior doctors, as using one-time e-mail addresses or ORCID submissions is suspect.
The phenomenon that a large number of retractions from Chinese hospitals in recent years is really worrying. Hospitals and medical libraries should take some measures to reduce the occurrences of research misconduct. In addition to the measure that seems to solve the problem fundamentally, but it is actually impossible to be promoted by individual hospitals: reform the doctor evaluation system, there are some short-term and mild measures that can be taken. In a prospective study on perceptions of Chinese Biomedical researchers towards research misconduct among Chinese biomedical researchers, more than 60% of participants thought that the punishment for research misconduct from authority and institutions was not severe enough (12). In fact, few hospitals in China have detailed rules and enforce them transparently, putting researchers with retracted publications at a disadvantage to their long-term interests, such as relegation, be limited to access to academic resources, and no promotion for several years (33). In 2019, the National Press and Publication Administration released the first industry standard against misconduct entitled Academic Publishing

Ethics approval and consent to participate
Not applicable.

Consent for publication
Not applicable.

Availability of data and materials
The datasets used and analyzed during the current study are available from the corresponding author on reasonable request.   status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.