Introduction

Background

Fraudulent research can have damaging consequences for public stakeholders, mislead subsequent research efforts that rely on false premises, and undermine public trust in academic research [1]. In medicine, the publication of untrustworthy research ultimately hinders evidence-based practice and can lead to patients receiving unsuitable treatments [2]. Medical research must transparently discredit false findings and stem the propagation of “research waste” [3]. Retraction is therefore an understudied mechanism within evidence-based medicine.

Retractions in medical research were first comprehensively examined by Budd et al. [4], who identified a total of 235 instances of retraction. Since that time, the speed of scholarly publishing has increased exponentially. Medical researchers and clinicians have increasingly been evaluated based on the quantity of publications they produce [5]. These trends coalesce to incentivize researchers to publish untrustworthy research. In response, the number of retractions in medical literature has grown exponentially over the past 25 years [6]. The large pool of available retraction data now allows for more nuanced analyses of various aspects of retracted medical literature. For example, large-scale retraction data have been used to analyze the specific medical research contexts of Brazil and Portugal [7]. However, retracted medical literature published by authors with a Taiwanese affiliation has not yet been systematically analyzed.

Taiwan, despite having a modest population of 23.4 million people, is the 21st largest contributor to research globally [8]. During the 21st century, academic misconduct allegations involving high-profile Taiwanese politicians have emerged in mainstream media. These include questioning the authenticity of former President Tsai Ing-Wen’s PhD [9], as well as allegations of misconduct against senior politicians. For example, a former Taiwanese Minister of Education Chiang Wei-Ling resigned from his position in 2014 after it was discovered that he contributed to several articles in a 60-paper peer review ring [10]. Despite significant public interest, independent Taiwanese investigators who examine potentially fraudulent publications have reported facing charges of extortion and attempts at character assassination [11]. Academic integrity and efforts to correct the academic record have been major topics within Taiwan’s political and academic realms, making Taiwan’s sociocultural relationship with retraction unique.

Aside from cases involving Taiwanese politicians, research examining specific instances of falsified and/or plagiarized data has also emerged. For example, Liu [12] presented evidence of 40 articles written by Taiwanese authors and published in international journals that contained questionable quantitative data. However, a broader analysis of retractions of Taiwanese-authored articles does not currently exist. Instead, large-scale analyses of publishing misconduct in Taiwan are often pooled with data from China [13]. Given that Taiwan’s research governance, academic career evaluation processes, and freedom of speech indices differ from those in China, it is not possible to infer insights into Taiwanese research integrity from China-centered analyses. Data on Taiwanese retractions have also been reported in the context of international comparisons with other countries. However, these data alone do not provide an adequate understanding of Taiwan’s retraction history, particularly within the medical disciplines.

Objectives

This study aimed to systematically examine the current state of retractions in Taiwanese medical research using data from the Retraction Watch Database (https://retractionwatch.com/) and PubMed. Specifically, the study counted the number of retracted articles with Taiwan-affiliated authors and compared these figures with those from other Asian countries, including China, Japan, Korea, and India. Additionally, the reasons for retraction were analyzed to determine whether any longitudinal trends exist in the nature of these retractions.

Methods

Ethics statement

This study did not involve human subjects; therefore, neither institutional review board approval nor informed consent was required.

Study design

This was a literature database-based observational study.

Data sources and measurement

After optimizing the data extraction and analysis protocols, a final copy of the Retraction Watch Database was downloaded on October 14, 2024. The data were filtered to include only articles with at least one author affiliated with Taiwan and classified as “(HSC) Medicine*”. Thus, the filtered dataset comprised only medical research articles with at least one Taiwan-affiliated author. To compare and contextualize retractions from Taiwan with those from other countries, general data on total research outputs were collected from SCImago Journal Rank (https://www.scimagojr.com/), along with data on medical research retractions from the Retraction Watch Database. A “retraction rate” for medical research was calculated for the 21 most productive countries (with Taiwan being the 21st largest producer of medical research publications) by dividing the total number of medical research retractions (identified through Retraction Watch Database searches and screening protocols identical to those described above) by the total national output of medical research. To analyze Taiwan-affiliated retractions in more detail, the dataset was supplemented with a PubMed search conducted on the same date for the term “Taiwan” in all fields, returning only results categorized as “retracted publication.” Duplicate entries and articles not related to medicine were removed.

This combined dataset was further analyzed using descriptive statistics and visualized with Microsoft Excel (Microsoft Corp) or RStudio (Posit PBC). The year of publication, the year of retraction, and the journal or publisher of publication were analyzed. To examine the rankings of the journals that published these retracted studies, data were extracted from the Web of Science. For any journal indexed in a Science Citation Index (including Emerging Sources Citation Index [ESCI], Social Sciences Citation Index [SSCI], and Science Citation Index Expanded [SCIE]), the ranking percentile for its primary Journal Citation Reports category was calculated. The number of times that other countries were affiliated with a retracted article was also counted and summarized.

Retraction can be too slow to adequately correct the scientific record before untrustworthy research findings are disseminated [14]. Therefore, the time interval between publication and retraction was examined for each article. Even if articles are retracted promptly, they may still be cited after retraction, highlighting the limitations of retractions in correcting the scientific record [4]. To examine how retracted medical research from Taiwan-based authors is cited, citation data were analyzed by searching for each retracted publication in the Scopus database. The citation breakdown for each article (excluding their respective retraction notices) was exported and triangulated with the article’s year of retraction. Citations accrued before or during the year of the article’s retraction were classified as pre-retraction citations. Citations accrued in the years following the retraction were classified as post-retraction citations.

Identifying specific reasons or causes for retraction can be very challenging, given that the retraction process is often shrouded in shame and dishonesty. Journals also vary widely in the level of detail provided in retraction notices. Consequently, no standardized inventory of retraction reasons exists for use in this analysis. Instead, the typology developed by Campos-Varela and Ruano-Raviña [15] was found to adequately characterize most of the retraction reasons in the dataset. Therefore, the retraction notice for each article (if available), along with any additional commentary on the retraction published by Retraction Watch, was reviewed and the retraction reason was classified into one of the eight categories adapted from Campos-Varela and Ruano-Raviña’s classification [15]:

(1) Unreliable data/conclusions within the paper: resulting either as a result of an unintentional error (scientific error, mistake on data processing or analysis) or as the result of an intentional manipulation, creation or fabrication of data or images.

(2) Authors: authorship disputes or vague statements made about disagreements regarding authorship.

(3) Plagiarism: text-recycling, redundant, or duplicated publications.

(4) Unethical research: no ethical approval or a lack of proper informed consent.

(5) Journal issues: editorial duplication of contents or an incorrect manuscript or version.

(6) Review process: fake peer review reports or other issues related with the peer review process.

(7) Other: a specific reason that does not fit into one of the above categories.

(8) Unknown: either no retraction statement can be located, or the retraction statement itself is too vague to identify a tangible reason for retraction.

Statistical analysis

Descriptive statistics were used to show the results. Where relevant, average results are reported as mean± standard deviation.

Results

Literature selection

The screening strategy used to generate the dataset for analyzing Taiwan medical retractions is presented as a flowchart adapted from the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines (Fig. 1) [16].

A total of 57,034 entries were retrieved from the Retraction Watch Database, of which 137 entries were listed under the “(HSC) Medicine” categories for Taiwan. To analyze Taiwan-specific retractions, results from the Retraction Watch Database were screened alongside returns from a PubMed search (Fig. 1). Ultimately, 137 retractions were identified from the Retraction Watch Database. Seventeen of these entries were excluded from analysis: six corrections, nine expressions of concern, one reinstatement of a previously retracted article, and one retraction of a preprint that has since been published in a peer-reviewed journal. This resulted in a dataset of 120 medical research publications produced by at least one author with a Taiwanese affiliation. An additional 61 eligible retractions were identified from the PubMed search and included, leading to a final dataset of 181 retractions. Research data for these 181 retracted articles are available in Dataset 1.

Rate of retracted articles by Taiwanese researchers

When compared to other research-productive nations, Taiwan’s medical literature has the 9th highest retraction rate out of 21 nations (Table 1).

Publication types and yearly amount of retracted articles

Of the 181 retracted publications, 158 (87.29%) were research articles, two (1.10%) were meta-analyses, seven (3.87%) were case reports, nine (4.97%) were review articles, and five (2.76%) were conference papers. The publication and subsequent retraction of these articles have increased since the earliest retracted article was published in 1992 (Fig. 2). Overall, there was an average delay of 3.14± 3.41 years between publication and retraction.

Analysis of international collaboration among retracted articles

The majority of the analyzed publications did not involve international collaboration. Specifically, 123 (67.96%) of the articles had solely Taiwanese affiliations, while the remaining 58 (32.04%) retracted articles were co-authored with researchers from at least one other country or region. Most of these collaborations were with the United States (n= 23, 39.66%) and China (n = 18, 31.03%). Korea, Pakistan, and Saudi Arabia each appeared in six publications (10.34%). India, Japan, and the United Kingdom each appeared in four publications (6.90%). Australia, Canada, and Egypt each appeared in three publications (5.17%). Colombia, France, Germany, Indonesia, Iran, Malaysia, Norway, and Vietnam each appeared in two publications (3.45%). Bangladesh, Cyprus, Hong Kong, Iraq, Kosovo, Lebanon, Macau, Myanmar, the Netherlands, Nigeria, Singapore, South Africa, Sweden, Thailand, Turkey, UAE, and Yemen each appeared in one publication (1.72%).

Publishers of retracted articles

The publishers of the articles in the dataset were diverse, with a total of 34 unique publishers identified. The four most common publishers were Elsevier (n=27, 14.92%), Hindawi (n=24, 13.26%), Springer (n= 23, 12.71%), and PLOS (n= 14, 7.73%). The publication venues were also varied (Tables 2, 3). Overall, the dataset comprised retracted literature from 119 peer-reviewed journals and one conference proceedings series (International Conference on Bioinformatics and Biomedical Engineering). Both international and local Taiwanese journals were represented.

A majority of the retracted studies were published in Web of Science-indexed journals (n= 171), except for those published in: Smart Materials in Medicine (n= 1), Medicina (n= 1), Key Engineering Materials (n= 1), Acta Anaesthesiologica Taiwanica (n = 1), Acta Paediatrica Taiwanica (n = 1), and the International Conference on Bioinformatics and Biomedical Engineering (n = 5). Nearly half of the retracted studies published in Web of Science-indexed journals were published in journals with a journal citation report ranking in the top quartile (Q1) for its category (Fig. 3).

Reasons for retraction

The reasons for retraction were identified according to the authors’ modified classification system originally developed by Campos-Varela and Ruano-Raviña [15]. Overall, 80 publications (44.20%) were retracted due to issues with data, 26 (14.36%) due to a compromised review process, 22 (12.15%) due to plagiarism, 8 (4.42%) due to authorship issues, 8 (4.42%) due to unethical practices, and 5 (2.76%) due to journal issues. One publication (0.55%) was retracted for another reason; in this case, the authors voluntarily withdrew their systematic review in order to publish an updated version with an expanded scope. The remaining 31 publications (17.13%) were retracted for unknown reasons. Overall, since 2006 there has been a decrease in the proportion of retractions due to data concerns and plagiarism, accompanied by an increase in those due to unethical research, authorship issues, and compromised review processes (Fig. 4).

Citation frequency of retracted articles

The retracted articles have been cited a total of 3,152 times, with each article receiving an average of 17.414 ± 28.97 citations. Only 145 (80.11%) of the retracted articles had at least one citation indexed by Scopus, and 1,073 of 3,152 citations (34.04%) occurred in the years following the articles’ retraction.

Discussion

Key results

Taiwan’s medical retraction dataset (n= 181) was derived from Retraction Watch and PubMed. Most of these articles (87.29%) were research papers, spanning 34 publishers and 119 venues. Retractions increased over a 33-year period, with an average delay of 3.14 years between publication and retraction. Data issues accounted for 44.20% of retractions, and nearly one-third of the articles involved international collaborations. Taiwan ranks 9th among the 21 most research-productive nations, with 6.08 retractions per 10,000 articles.

Interpretation/comparison with previous studies

Analyzing retracted literature provides insights into research practices that contribute to untrustworthy work and reveals how a research community upholds academic integrity. Taiwan’s retraction rate in medical disciplines is lower than that of other Asian countries such as China, India, Korea, and Japan (Table 1). Nonetheless, it still ranks 9th among the top 21 producers of medical research. Furthermore, this analysis shows that the publication of untrustworthy research by Taiwanese authors has increased over the past two decades (Fig. 2), although this growth in retractions is relatively lower than global publishing trends [17]. It is therefore important to better understand how Taiwanese medical research cultures intersect with academic integrity.

The high rate of collaboration with the United States–affiliated authors among the retracted literature is unsurprising, given that approximately 70% of Taiwan’s international clinical research collaborations involve the United States [18]. In contrast, the rate of international collaboration with Chinese authors was previously estimated at only 4.69% [18]. The disproportionate representation of Chinese co-authors in the analyzed dataset (31.03% of all papers with international collaboration) may be attributed either to a recent increase in collaboration between Taiwan-based researchers and Chinese authors or to the high retraction rate of China-affiliated medical research.

The longitudinal changes in retraction reasons from 2005 to 2024 (Fig. 4) likely reflect broader shifts in the publishing industry. Plagiarism and data manipulation—such as image cropping, splicing, or numerical fabrication—can now be detected automatically through screening tools. These tools, along with high-profile dismissals of Taiwanese researchers repeatedly found to manipulate data [19], may be deterring current authors from engaging in such practices. However, the proportional decrease in retractions for these reasons may also indicate a reduced interest by editorial teams in following up on expressions of concern amid an accelerated publishing context [14].

Conversely, an increasing number of Taiwanese medical researchers have had their findings retracted due to compromised peer review. This may take the form of “peer review rings,” where authors unethically approve each other’s work, or through participation in commercial paper mills that collaborate with journal editors to bypass proper peer review. Even authors with honest and unmanipulated work may have their articles retracted as part of broader crackdowns on journals or publishers that fail to maintain peer review standards. These emerging issues may be difficult for Taiwanese scholars to detect, as they rarely discuss predatory publishing with their librarians [20]. This analysis underscores the importance of informing Taiwanese medical researchers about emerging misconduct behaviors in scholarly publishing so they can identify trustworthy venues for their work.

Although no clear trends were observed in the publication venues of these retracted articles, it is encouraging that local Taiwanese journals have also retracted articles. This demonstrates a baseline respect for academic integrity among Taiwanese medical research editorial teams. The majority of journals that have retracted medical research by Taiwanese authors are highly ranked (Fig. 3). This may be because highly cited journals, with larger readerships, have a higher chance of detecting retraction-worthy flaws. Based on this logic, many untrustworthy publications in lower-ranking journals may still await detection.

Limitations

The total volume of biomedical research varies depending on the literature databases selected. Furthermore, some local journals published in Taiwan that are not indexed in international databases were not included in this study; the same applies to other countries. Therefore, the retraction rate for each country may not accurately reflect the rate of retraction for local publications.

Conclusions

The rate of retractions in Taiwanese medical research has increased over the past two decades, largely due to a growing number of authors bypassing rigorous peer review. However, compared to broader global trends, Taiwan has succeeded in slowing the growth of retracted literature. Overall, this study underscores the importance of supporting research integrity training among Taiwanese medical researchers. The analysis also predicts a continued increase in the publication of untrustworthy research, highlighting the need for improved editorial vigilance. Future research—such as interviewing Taiwanese authors and editors about their awareness and attitudes toward questionable research practices—may further enhance research integrity.

Notes

Conflict of Interest

No potential conflict of interest relevant to this article was reported.

Funding

The author received no financial support for this article.

Data Availability

Dataset file is available from the Harvard Dataverse at https://doi.org/10.7910/DVN/Z0I14Q.

Dataset 1. Research data of 181 retracted articles by Taiwanese authors in biomedical journals.

kcse-360-dataset-1.xlsx

Supplementary materials

Fig. 1.

Screening strategy to generate the dataset used to analyze Taiwan medical retractions. RWD, Retraction Watch Database.

Fig. 2.

Distribution of publication and retraction dates. ^a)Only includes data from January 1, 2024 to October 14, 2024.

Fig. 3.

Box plot of the Web of Science-indexed publications that were later retracted. The Y-axis represents the percentile ranking of the journal’s 2023 Journal Citation Reports ranking for the first category listed (for example, publications located between 0% and 25% on the Y-axis were published in what is currently a Q1 journal).

Fig. 4.

Longitudinal analysis of reasons for retraction.

References

• 1. Kennedy D. Reflections on a retraction. Science 2000;289:1137. https://doi.org/10.1126/science.289.5482.1137. ArticlePubMed
• 2. Steen RG. Retractions in the medical literature: how many patients are put at risk by flawed research? J Med Ethics 2011;37:688-92.https://doi.org/10.1136/jme.2011.043133. ArticlePubMed
• 3. Grimes DR. Is biomedical research self-correcting? Modelling insights on the persistence of spurious science. R Soc Open Sci 2024;11:231056. https://doi.org/10.1098/rsos.231056. ArticlePubMedPMC
• 4. Budd JM, Sievert M, Schultz TR, Scoville C. Effects of article retraction on citation and practice in medicine. Bull Med Libr Assoc 1999;87:437-43.https://pubmed.ncbi.nlm.nih.gov/10550028/. PubMedPMC
• 5. Hong J, Xu J, Sun X. Young doctors and the pressure of publication. Lancet 2013;381:e10-1.https://doi.org/10.1016/S0140-6736(13)60934-4. Article
• 6. Oransky I, Fremes SE, Kurlansky P, Gaudino M. Retractions in medicine: the tip of the iceberg. Eur Heart J 2021;42:4205-6.https://doi.org/10.1093/eurheartj/ehab398. ArticlePubMed
• 7. Candal-Pedreira C, Ruano-Ravina A, Rey-Brandariz J, et al. Evolution and characterization of health sciences paper retractions in Brazil and Portugal. Account Res 2023;30:725-42.https://doi.org/10.1080/08989621.2022.2080549. ArticlePubMed
• 8. SCImago. SCImago Journal & Country Rank [Internet]. Scimago; [cited 2024 Oct 4]. Available from: https://www.scimagojr.com/countryrank.php.
• 9. Liu JC. Ing-Wen Tsai’s 1983 LSE PhD thesis: an annotation. J Sci Humanit 2020;7:1-267.https://doi.org/10.6535/JSH.202002_7(2).0001. Article
• 10. Department of Information Services. Resignation of Minister of Education Chiang Wei-ling accepted [Internet]. Executive Yuan; 2014 [cited 2024 Sep 28]. Available from: https://english.ey.gov.tw/Page/61BF20C3E89B856/8fb77e40-9cc1-4f3f-b57c-1be29e03d41e.
• 11. Liu JC. Is an academic cockroach making profit from disclosing defects in scholars’ papers? J Sci Humanit 2019;6:1-53.https://doi.org/10.6535/JSH201908_6(4).0001. Article
• 12. Liu JC. Comment on 40 Taiwanese scholars’ papers published in international journals with common numerical research results. J Sci Humanit 2021;8:80-237.https://doi.org/10.6535/JSH.202111_9(1).0005. Article
• 13. Zhang J. Reasons and countermeasures for retraction articles in the field of neuroscience in China from 2010 to 2021. J Cap Med Univ 2022;43:618-21.https://doi.org/10.3969/j.issn.1006-7795.2022.04.016. Article
• 14. Besançon L, Bik E, Heathers J, Meyerowitz-Katz G. Correction of scientific literature: too little, too late! PLoS Biol 2022;20:e3001572. https://doi.org/10.1371/journal.pbio.3001572. ArticlePubMedPMC
• 15. Campos-Varela I, Ruano-Raviña A. Misconduct as the main cause for retraction. A descriptive study of retracted publications and their authors. Gac Sanit 2019;33:356-60.https://doi.org/10.1016/j.gaceta.2018.01.009. ArticlePubMed
• 16. Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ 2021;372:n71. https://doi.org/10.1136/bmj.n71. ArticlePubMedPMC
• 17. Pourhoseingholi MA, Rahimibashar F, Khosh Fetrat M, Vahedian-Azimi A. Is the ratio of retracted iranian papers compatible with their research growth rate? An international database survey. Hosp Pract Res 2021;6:29-34.https://doi.org/10.34172/hpr.2021.06. Article
• 18. Chen TJ, Chen YC, Hwang SJ, Chou LF. International collaboration of clinical medicine research in Taiwan, 1990- 2004: a bibliometric analysis. J Chin Med Assoc 2007;70:110-6.https://doi.org/10.1016/S1726-4901(09)70340-5. ArticlePubMed
• 19. Marcus A. Researcher formerly of OSU and Taiwan’s Academia Sinica gets 10-year ban [Internet]. Retraction Watch; 2020 [cited 2024 Sep 27]. Available from: https://retractionwatch.com/2020/02/25/researcher-formerly-of-osu-andtaiwans-academia-sinica-gets-10-year-ban/.
• 20. Ke HR. A study on Taiwanese scholars’ perceptions and practices of open access journals. J Educ Media Libr Sci 2022;59:239-76.https://doi.org/10.6120/JoEMLS.202211_59(3).0022.RS.AM. Article

Figure & Data

References

Citations

Citations to this article as recorded by