key: cord-0888825-l5hcnlf5
authors: Cortegiani, Andrea; Ippolito, Mariachiara; Ingoglia, Giulia; Manca, Andrea; Cugusi, Lucia; Severin, Anna; Strinzel, Michaela; Panzarella, Vera; Campisi, Giuseppina; Manoj, Lalu; Gregoretti, Cesare; Einav, Sharon; Moher, David; Giarratano, Antonino
title: Citations and metrics of journals discontinued from Scopus for publication concerns: the GhoS(t)copus Project
date: 2020-08-26
journal: F1000Res
DOI: 10.12688/f1000research.23847.2
sha: 95cece38f01df946a5f8447e4ef255754b621b45
doc_id: 888825
cord_uid: l5hcnlf5

Background: Scopus is a leading bibliometric database. It contains a large part of the articles cited in peer-reviewed publications . The journals included in Scopus are periodically re-evaluated to ensure they meet indexing criteria and some journals might be discontinued for 'publication concerns'. Previously published articles may remain indexed and can be cited. Their metrics have yet to be studied. This study aimed to evaluate the main features and metrics of journals discontinued from Scopus for publication concerns, before and after their discontinuation, and to determine the extent of predatory journals among the discontinued journals. Methods: We surveyed the list of discontinued journals from Scopus (July 2019). Data regarding metrics, citations and indexing were extracted from Scopus or other scientific databases, for the journals discontinued for publication concerns. Results: A total of 317 journals were evaluated. Ninety-three percent of the journals (294/317) declared they published using an Open Access model. The subject areas with the greatest number of discontinued journals were Medicine (52/317; 16%), Agriculture and Biological Science (34/317; 11%), and Pharmacology, Toxicology and Pharmaceutics (31/317; 10%). The mean number of citations per year after discontinuation was significantly higher than before (median of difference 16.89 citations, p<0.0001), and so was the number of citations per document (median of difference 0.42 citations, p<0.0001). Twenty-two percent (72/317) were included in the Cabell’s blacklist. The DOAJ currently included only 9 journals while 61 were previously included and discontinued, most for 'suspected editorial misconduct by the publisher'. Conclusions: Journals discontinued for 'publication concerns' continue to be cited despite discontinuation and predatory behaviour seemed common. These citations may influence scholars’ metrics prompting artificial career advancements, bonus systems and promotion. Countermeasures should be taken urgently to ensure the reliability of Scopus metrics for the purpose of scientific assessment of scholarly publishing at both journal- and author-level.

Scopus is a leading bibliometric database launched in 2004 by the publishing and analytics company Elsevier. It was developed by research institutions, researchers and librarians, and contains the largest number of abstracts and articles cited in peer reviewed academic journal articles that cover scientific, technical, medical, and social science fields 1 .

Scopus provides bibliometric indicators that many institutions use to rank journals to evaluate the track record of scholars who seek hiring or promotion. These metrics are also used to allocate financial bonuses or to evaluate funding applications [2] [3] [4] . Ensuring the quality of the content of the Scopus database is therefore of great importance.

Scopus indexed journals undergo evaluation and periodic review by an independent and international Content Selection and Advisory Board (CSAB), a group of scientists, researchers and librarians, comprised of 17 Subject Chairs, each representing a specific subject field-and by a computerized algorithm 1 . At any time after journal inclusion, concerns regarding its quality may be raised by a formal complaint, thereby flagging the journal for re-evaluation by the CSAB. Should the CSAB panel determine that the journal no longer meets Scopus standards, new articles from that journal are no longer indexed 1 . One of the most common reasons for discontinuation is 'publication concerns', which refers to the quality of editorial practices or other issues that have an impact on its suitability for continued coverage 5 . The list of the discontinued sources is publicly available and is updated approximately every six months 6 . However, articles published in journals that were discontinued and are no longer indexed, are probably not removed from the Scopus database.

It has been claimed that a number of journals discontinued from Scopus for publication concerns might be so-called 'predatory' journals 5 . Predatory journals "prioritize self-interest at the expense of scholarship and are characterized by false or misleading information, deviation from best editorial and publication practices, a lack of transparency, and/or the use of aggressive and indiscriminate solicitation practices" 7 . Since researchers are pressured to publish in indexed journals, predatory journals are constantly trying to be indexed in the Scopus database, thereby boosting their attractiveness to researchers 2, 8 . Having articles from predatory journals indexed in Scopus poses a threat to the credibility of science and might cause harm particularly in fields where practitioners rely on empirical evidence in the form of indexed journal articles 8,9 .

We hypothesize that, even though Scopus coverage is halted for discontinued journals, they are still cited, as all their documents, that are already indexed, remain available to users. To date, the metrics of those journals discontinued for publication concerns have not been studied yet. Therefore, in the present analysis we set out to (1) evaluate the main scientific features and citation metrics of journals discontinued from Scopus for publication concerns, before and after discontinuation, and (2) determine the extent of predatory journals included in the discontinued journals.

The freely accessible and regularly updated Elsevier list (see Source data) of journals discontinued from the Scopus database (version July 2019) 10 was accessed on 24 th January 2020 (See Underlying data 11 ). We restricted our analysis to journals discontinued for "publication concerns". Journals were checked for relevant data (described below), which were then independently collected by four pairs of authors (MI and GI, AM and LC, AS and MS, VP and AC), each pair being assigned one quarter of the data to be collected in duplicate. The data were collected using a standardized data extraction form (Underlying data Table 1 ). A second check to confirm the data and resolve discrepancies was performed by four additional authors that had not been involved in data collection (LM, CG, SE, AG). Data collection was initiated on 24 th January and completed by the end of February 2020. Confirmed data were registered on an Excel datasheet (Underlying data, Table 1 

Retrieved data and sources Data were extracted either from the Scopus database 10 or by searching other sources, such as SCImago Journal & Country Rank (SJCR) 12 , Journal Citation Reports 13 , Centre for Science and Technology Studies (CWTS) Journal Indicators 14 , Beall's updated List 15 , Directory of Open Access Journals (DOAJ) 16 , PubMed 17 and Web of Science 18 . Open Access policy was checked on journals websites. The standardized data extraction form, independently applied by eight authors (MI, GI, AM, LC, AS, MS, VP, AC), was used to collect the following data: journal title, name and country of the publisher, the number of years of Scopus coverage, year of Scopus discontinuation, subject areas and sub-subject areas, Impact Factor (IF), CiteScore,

We are glad to submit a new version of our manuscript, now entitled "Citations and metrics of journals discontinued from Scopus for publication concerns: the GhoS(t)copus Project", and incorporating the insights provided by the reviewers.

In this revised version, more words of caution have been inserted regarding the interpretation of our findings, and the lack of a control group has been listed among the limitations of the study.

Typos were amended and few data were corrected. The findings and conclusions remain consistent with the previous version since no substantial changes were made. English form was revised.

More details have been added in the methods section, in order to improve the reproducibility of the research.

Furthermore, some minor changes were made in the tables, the figure caption and the reference list, following reviewers' suggestions.

Peer review had an important role in improving this manuscript, that now results more balanced. Also, new insights for further research questions have been included.

SCImago Journal Rank (SJR), Source Normalized Impact per Paper (SNIP), best SCImago quartile, the indexing of at least one article in PubMed, Web Of Science (WOS) and DOAJ (for open access journals) indexing, presence in the updated Beall's List, total number of published documents and total number of citations. All the metrics were checked on the year of Scopus discontinuation. In cases of discrepancies between Scopus data and other sources, Scopus data was preferred.

We defined the 'before discontinuation' time frame as the period included within the first year of journal coverage by Scopus and the year of discontinuation, which was not included in our calculations. The 'after discontinuation' time frame, was defined as the period included within the year of Scopus discontinuation and 2020. If the journal had been discontinued more than once, the time frame was based on the last one, according to the date of the last document displayed in the Scopus database. Citations 'before' and 'after' the date of discontinuation were manually counted based on either the Scopus journal overview or the downloadable tables made available by Scopus upon request (see Source data). When evaluating the presence of articles in PubMed (e.g. PubMed Central) and WOS and DOAJ indexing, 2019 was considered the reference year, preventing disadvantages for journals with time gaps for publication.

We calculated the median number of cumulative citations across all discontinued journals per year of coverage and defined it as 'Citations per year'. We also calculated the median number of cumulative citations across all discontinued journals per document ('Citations per document'). We included all documents indexed in Scopus, regardless of type. Finally, one author (AS) checked whether discontinued journals were present in Cabell's whitelist or blacklist 19 or the DOAJ's list of discontinued journals 20 . As some of the journals included in the blacklist lack ISSNs or other unique identifiers, the comparison of the three lists with Scopus's discontinued journals was based on matching the journals' names by similarity using the Jaro-Winkler algorithm in RStudio Desktop 1.2.5033 and RecordLinkage 0.4-11.2 following the approach developed by Strinzel et al. (2019) 21,22 . The Jaro-Winkler metric, scaled between 0 (no similarity) and 1 (exact match), was calculated for all possible journals' pairings 23 . We manually inspected all pairs with a Jaro-Winkler metric smaller than one in order to include cases where, due to the orthographical differences between the lists, no exact match was found. For each matched pair, we compared journal publishers and, where possible, ISSNs in order to exclude cases where two journals had the same or a similar name but were edited by different publishers.

Full definitions and descriptions of the sources and metrics are reported in the Extended Data Appendix 1 24 .

All data management and calculations were performed using Microsoft Excel (version 2013, Microsoft Corporation®, USA) and GraphPad Prism (version 8.3.1, 322, GraphPad software®, San Diego California). Variable distribution was assessed for normality using the D'Agostino-Pearson test. For variables with normal distribution means and standard deviations (SDs) were reported. For non-normally distributed data medians, interquartile ranges (IQRs, 25th-75th) and ranges (minimum value -maximum value) were reported. Categorical data were expressed as proportions and percentages.

The paired sample t test or the Wilcoxon matched-pairs signed ranked test were used to compare journal data before and after Scopus discontinuation, as appropriate.

Data could be retrieved regarding 317 of the 348 journals listed as discontinued (91.1%). The remaining journals were not found on the Scopus database using the search tool.

Among the 135 publishers identified, those with the largest number of discontinued journals were: Academic Journals Inc. (39/317; 12.3%), Asian Network for Scientific Information (19/317; 6.0%), and OMICS Publishing Group (18/317; 5.7%). Table 1 reports the distribution of journals discontinued from Scopus by publisher. United States (76/317, 24%), India (63/317, 20%) and Pakistan (49/317, 15%) were countries most commonly declared as publisher headquarters ( Figure 1 and Table 2 ).

The subject areas with the greatest number of discontinued journals were Medicine (52/317; 16%), Agriculture and Biological Science (34/317; 11%), and Pharmacology, Toxicology and Pharmaceutics (31/317; 10%) Table 3 and Extended data  Table 1 25 report the distribution of discontinued journals by subject area and sub-area in full. Of these journals, 93% (294/317) declared they published using an Open Access model.

First subject area as displayed in Scopus. Note: a journal may have more than one subject area. Table 4 shows the characteristics and metrics of the journals at the time of their discontinuation.

The median time of Scopus coverage prior to discontinuation of the journals was 8 years (IQR 6-10, range 1-54). In total, 299 journals had been assigned to a SCImago quartile (Q); 39 of them (13%) listed in Q1 or Q2, and 260 in Q3 or Q4 (87%). Only ten of the discontinued journals had an Impact Factor at the year of discontinuation, with a median value of 0.84 (IQR 0.37-2.29, range 0.28-4). Table 5 shows the total number of documents and citations, the total number of documents per journal and the citations count before and after Scopus discontinuation. The total number of citations received after discontinuation was 607,261, with a median of 713 citations (IQR 254-2,056, range 0-19,468) per journal.

Paired t-tests (Wilcoxon matched-pairs signed rank test) revealed that the number of citations per year after discontinuation was significantly higher than before (median of difference 16.89 citations [-13.68-117.5] (-1427-3491), p<0.0001). Likewise, the number of citations per document proved significantly Indexing in Cabell's lists, updated Beall's list, DOAJ and scientific databases Among the discontinued journal, 22% (72/317) were included in the Cabell's blacklist, while 29 (9%) were currently under review for inclusion. Only five journals (2%) were included in Cabell's whitelist. In 243 cases (76.6%), either the journal publisher was included in the updated Beall's list of predatory publishers or the journal was included in the corresponding list of standalone journals (76.6%). The DOAJ currently includes only 9 journals. In total, 61 journals were previously included and discontinued by DOAJ; in 36 cases the reason was 'suspected editorial misconduct by the publisher' in 23 instances it was 'journal not adhering to best practice' and in one case 'no open access or license info'. Table 6 shows the indexing in Web of Science, updated Beall's list, Cabell's white-and blacklist, and DOAJ (both included and discontinued) and the presence of articles in PubMed. 

The present study aimed to scrutinize the main features of journals whose coverage was discontinued by Scopus due to publication concerns. To do so, (a) we counted and compared citation metrics per journal and per document obtained before and after discontinuation, and (b) we accessed established blacklists and whitelists dealing with the issue of predatory publishing, i.e. Cabell's and updated Beall's list, as well as the DOAJ.

Our main finding was that articles published in these journals before discontinuation remain available to users and continue to be cited after discontinuation, and even more so than before. Moreover, a large number of the discontinued journals are likely to be predatory.

A previous analysis conducted to evaluate the scientific impact of predatory publishing has concluded that "articles published in predatory journals have little scientific impact" 26 . The study evaluated Google Scholar and Scopus citation statistics of 250 randomly sampled articles, that have been published in predatory journals in 2014. The citations were then compared to those of a control group of articles, published in journals included in Scopus database. Our study aimed to evaluate and describe the metrics and citations of all the journals discontinued from Scopus for 'publication concerns'. At a secondary stage, the presence of these journals in the Cabells' and Beall's lists was investigated. The different purposes and designs of the two studies may explain the different findings.

Although Scopus rigorously controls content quality and warns users when a journal is discontinued in its source details, the average user rarely accesses journaldetails, usually focusing on article contents alone. As a result the reader remains unaware that the article they have accessed was issued by a journal discontinued for publication concerns. Therefore, articles issued by journals whose scientific reputation is currently deemed questionable continue to be cited as content from legitimate, up-to-standard journals. Quantification of the effect of discontinuation on the likelihood of citation shows that the articles published by these journals received significantly more citations after discontinuation than before.

Apart from dangerous exposure of scholars, clinicians and even patients to potentially dubious or low quality contents, citations from discontinued journals pose a serious threat to assessment of scientific merit and quality by institutions and academia. These citations contribute to the calculation of author metrics by Scopus. Among these metrics is included the Hirsch index (H-index) 27 , a lead descriptor of productivity and scientific impact, upon which career advancements are often determined [2] [3] [4] . The fact that discontinued journals contribute to academic promotion is a pertinent issue, and has inspired the vignette depicted in Figure 2 : discontinued journals may inflate authors' metrics lifting them unnaturally and effortlessly.

Of greatest concern is our finding that many of the discontinued journals display predatory behaviors in claiming to be open access, without actually being indexed in DOAJ.

Exploitation of the open-access publishing model has been shown to go hand in hand with deviation from best editorial and publication practices for self-interest 7 . Predatory journals are not only associated with poor editorial quality, but are also deceptive and misleading by nature, i.e. they prioritize self-interest at the expense of scholars, and lack transparent and independent peer review 7,28 . Young researchers from lowand middle-income countries are probably most susceptible to the false promises and detrimental practices of predatory journals. However, "predatory scholars" also seem to exist, possibly sharing a common interest with deceptive journals and publishers, knowingly using them to achieve their own ends 29,30 .

The policy underlying the decision to keep publications prior to discontinuation of indexing is clear. Some of these publications may actually fulfill publishing criteria (e.g. International Committee of Medical Journal Editors, Committee on Publication Ethics). It would be unfair to punish researchers for an eventual deterioration in journal performance; changes in the standards employed by the journal may change over time and the researchers may be unaware of quality issues. On the other hand, as the integrity of the editorial process cannot be vouched for, it is ethically untenable to keep such data available without clearer warnings.

One measure that could be undertaken immediately is, for example, flagging of articles that have been published in discontinued journals with clearly visible information regarding journal discontinuation, its date and its cause. Submitting articles published a certain amount of time before journal discontinuation to post-publication open peer-review is also a possibility. However, as solutions to this problem must balance fairness towards publishing researchers with ensuring the correctness of the metrics and citations deriving from these journals, Scopus may need to to set criteria for deleting discontinued journals from the publicly available database or, in the least, stop tracking their citations. Such measures must only be applied by the CSAB case-by-case, after evaluating the full impact of such action and the severity of the potential misconducts. At the author-level, an alternative may be the provision of two metrics: one with and one without citations from publications in discontinued journals.

This analysis is not free of limitations. First, this study lacks a control group of journals whose coverage had not been discontinued in the Scopus database. Therefore the differences we identified in the number of citations before and after discontinuation require further validation. Second, we included the year of discontinuation in the "after discontinuation" period, starting from January 1 st . This decision may have led to some overestimation in the number of citations received after discontinuation. Third, we included only those journals discontinued from Scopus for "publication concerns" but were not able to retrieve details regarding the specific concern raised. Finally, we did not evaluate the impact of the citations received after discontinuation on author-level metrics.

Journals whose coverage in Scopus has been halted for publication concerns continue to be cited. This paradox may influence scholar metrics, potentially prompting career advancements and promotions. Further studies are needed, also investigating the journals discontinued from Scopus using the criteria "outlier performance -radar", particularly effective in flagging potential predatory journals. Countermeasures should be taken to ensure the validity and reliability of Scopus metrics for both journals and authors due to their importance for scientific assessment of scholarly publishing. Creative thinking is required to resolve this issue without punishing authors who have inadvertently published good quality papers in a failing or predatory discontinued journal. I like to thank the authors and the journal for considering me as a peer for this review. I very much enjoyed this revised paper and certainly the indebt revision of the comments from the first round. I only have few general comments on the external validity of this paper:

Exclusion of journals only based on publication concerns may not reflect the complexity facing us of information overload.

The proxy control group, which is absent and is mentioned by the authors is in reality a comparison with journals using an editorial system based on peer review as the gold standard that ensures adequate quality. But considering the issues surrounding the famous hydroxychloroquine paper in Lancet and the change of their editorial policy this week (not published in detail how this will affect the peer review system), the complexity of this issue becomes even more apparent.

I personally am far from convinced that the current editorial system and peer review is at all adequate or up-to-date to detect scientific fraud and ensure high quality papers. What we see is an inflation in the number of journals opting for a payment system. Thus, the financial incentives for publication and the consequent pressure on editors is increasing. This is not really addressed in this paper. Additionally, the journals are in essence cherry picking. Peer reviewers do the job, researchers do their part and the funding comes from public or private sources. The journals end up making the money. And there is often very little transparency about the quality of editorial system and the peer review process in even very well-established papers. For instance, one never knows the number of peers, name, affiliation, conflict of interest and the extent of data scrutiny. Finally, often data are not provided, shared and even to a lesser degree re-analyzed unless very controversial or with a high clinical impact. And there is often very little effort nor incentive to opt for reproducibility.

What we are witnessed to is a tsunami of useless scientific papers. For instance only 3 percent of systematic reviews published today have adequate quality and address the issue of random error and reproducibility. Thus, a major limitation of this study is that it only reflects the general quality of journals as a proxy indicator for scientific malpractice and retraction per se.

But overall, an enlightening work that adds valuable information to complexity of the issue of predatory journals and their impact.

Is the study design appropriate and is the work technically sound? Yes

Are all the source data underlying the results available to ensure full reproducibility? Yes

Competing Interests: No competing interests were disclosed. 

Library of the University of Geneva, Geneva, Switzerland

Library of the University of Geneva, Geneva, Switzerland

Division of Anaesthesiology, Department of Acute Medicine, Geneva University Hospitals, Institute of Global Health, Faculty of Medicine, University of Geneva, Geneva, Switzerland

Thank you for giving us the opportunity to read this article in which the authors describe the characteristics, citations and metrics of journals that have been indexed in the Scopus database, at some point, and have afterwards been "discontinued" within Scopus for different reasons, summarised as "publication concerns".

Since the articles that have been published before the journal's indexation was discontinued remain in the database, and can still be found, they may still be cited. The authors find this to be particularly problematic since they believe that these journals may be what are often called "predatory journals", and therefore may threaten the credibility of science, by polluting the database with "weak research". Therefore, the authors aimed to compare the number of citations per year, and per journal, and per document, before and after the journal was delisted. They conclude that the number of citations was actually higher after the journal was discontinued from Scopus. Although we understand the problem these authors try to highlight, we have some major concerns regarding some aspects of this study (mainly related to baseline assumptions and lack of clear definition) and also some minor ones.

Baseline assumption: In this article, the authors suggest that if a journal is being indexed, even for a long period of time (half of them have been indexed for 8 to 54 years), and is encountering "publication concern", then all the previously published articles should become suspicious of bad science. We are not sure this should be considered straight forward, for the reason developed under our second major concern.

"Predatory journals": the problem of the lack of a clear definition of what a predatory journal is, remains. The authors use different sources to try to identify journals as "predatory" and we can only realise that the sources do not seem to agree. Although authors auto cite their own "consensus definition" of predatory journals and publishers "(..) entities that prioritize self-interest at the expense of scholarship and are characterized by false or misleading information, deviation from best editorial and publication practices, a lack of transparency, and/or the use of aggressive and indiscriminate solicitation practices." they fail to underline that not everybody agrees with this definition. Also, the recent COVID-19 debacle of very low-quality scientific publications, published in usually highly regarded journals, suggests that bad peer-review and misleading articles may not be a characteristic of any journal. Also, it remains unclear to us how a journal may be indexed for 8 to 10 years, and all of a sudden become "predatory". Or was it predatory in the first place, but was only uncovered after such a long time? If this is what the authors suggest, then what should we think about "recently" indexed journals? They may all be predatory as well, and will only be uncovered in 5 to 10 years?

"Publication concerns": This term needs to be better defined in order to really understand what lies behind it. It remains unclear why these journals have been excluded from the Scopus database at some point. Interestingly, half of these journals have been deemed good enough to figure in the database for more than 8 years… that's a lot! And all of a sudden, they are not judged acceptable anymore and are discontinued from Scopus. Ok, why not. It may take some time before someone alerts Scopus of the misbehaviour of a given journal, although more than 10 years for 25% of them seems a lot. Or could it be a problem behind the vague concept of "publication concerns"? Could it be that the publication has stopped? Or the journal has changed its name? Or has merged with another one? Or has changed in quality over time? Illustrating some of the reason for discontinuation would help the reader understand the context.

According to Scopus' document cited in the article (ref. 5), there are 3 causes prompting Scopus to launch a journal re-evaluation: Under performance -metrics ; Outlier performance -radar ; Publication concerns. It might have been interesting to analyse the journals removed using the criteria "outlier performance -radar" as well as, according to Scopus document (ref. 5 cited by the authors) it is "particularly effective in flagging potential predatory journals." Scopus describes it as "an algorithm that flags journals based on approximately 40 outlier predictors, including sudden change in output volume, sudden change in publishing country and/or affiliations, and high journal/author self-citation rates."

The authors are worried that the citations of these journal have increased after the journal's indexation was discontinued in the Scopus database. The problem here is that they do not seems to consider the fact that this may be the case for all journals (those indexed and those discontinued) which is probably due to the rapid increase in the number of publications over time. Unfortunately, this study lacks a "control group" (journals whose coverage has NOT been discontinued in the Scopus database) which could have help the reader understand whether the increase in citation of these journals was similar, was higher, or was lower than that of "legitimate journals".

Underlying discourse: The term "inflated" used in the title, in Figure 2 and conclusion suggests manipulation or distortion of citations and an artificial advantage for authors of articles published in predatory journals before they are removed from Scopus. This is not demonstrated by the reasoning and data used in the article as a basis for comparison is missing.

Methods and reproducibility: While the authors have provided data alongside the article, we have not been able to reproduce some of their results, such as "citations per year" presented in table 5. Data presented in "underlying data table 1" would benefit from better variable descriptions, such as where exactly was the information collected from, and the date of its collection. Some variable names and analysis are misleading, such as "Actual Pubmed", described in methods section as "inclusion in PubMed" and in table 6 as "main database indexing". It does not reflect whether the journal is currently indexed in PubMed, but may in some cases only indicate that a single article is present in PubMed or selected citations, due to their deposit in PMC (eg. "Advanced Materials Letters"). Some data seem a bit bizarre… and information provided by the authors like "Citation before and after the date of discontinuation were manually counted based on either the Scopus journal overview or the downloadable tables made available by Scopus upon request (see source data)" (p.3) did not allow us to double check some numbers that were weirdly extreme, and potential typos. Some counts of the number of citations seem erroneous, leading to an aberrant number of citations per document for journals like "Mental Health in Family Medicine" (80 citations per document before discontinuation) or "Pharmacognosy Reviews" (170 after). Other example of bizarre data: according to "underlying data table 1" the journal "Advanced material research" has been indexed for 10 years (from 2004 to 2014) and has received during this period only 3 citations. However, after having been delisted from the Scopus database, during a 6 year period (2014 to 2019), it has received 13875 citations. Any thoughts on how/why this could have happened?

Minor concerns:

Background: "contains the largest number of abstract and articles…" -> "One of the largest" could be better, some databases are bigger than Scopus ( Methods: The use of the term "discontinued" both for DOAJ (Results) and for journal publication (Background) is confusing. Should we say "excluded" or "delisted"? ○ Results: "317 journals were evaluated" but next sentence states ninety-three percent of the journals (294/318)" -> typo for 318? ○ Results: "the mean number of citations per year after discontinuation was significantly higher than before, and so was the number of citation per document". Unclear whether the median difference of 64 is per journal, or cumulative across all "discontinued" journals? What are "documents"? Do you mean "articles"? or are there any other types of publication? ○ Conclusions: it's unclear how the conclusion regarding "predatory journals" is drawn. Also, we don't think the career advancement are "artificial", they are real! Although maybe "undue"? ○ Introduction: "publications from no longer indexed journals may not be removed retrospectively … hence articles … could remain part of the database 7 " p.3 -> this conditional statement seems to contradict abstract which categorically states "These journals remains indexed" as well as the author's conclusion "we propose that CSAB could apply these measure case-by-case". Reference 7, linked with statement, was not helpful to clarify. Citations by year in table 5: The number of years before and after the journal is removed from Scopus is very different, the average is more than 9 years before and 4 years after (median of 8 and 4 respectively) which makes the comparison in Table 5 Publishing in a "predatory journal" may be one of them, but auto-citation is also one. Of the 30 references cited at the end of this paper, 11 (37%) are auto-citations (citation of a reference including at least on author of the present paper), 7 (23%) are articles from others, and the remaining 12 were websites.

In Underlying data table 1: last column title -DOAJ instead of DOAH ? ○ p.3 §3 "journal are no longer be indexed" -> "journal are no longer indexed" ○ p.3 §5 "still they can get" -> "they can still get" ○

Are all the source data underlying the results available to ensure full reproducibility? Yes

Competing Interests: No competing interests were disclosed.

We confirm that we have read this submission and believe that we have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however we have significant reservations, as outlined above. Since the articles that have been published before the journal's indexation was discontinued remain in the database, and can still be found, they may still be cited. The authors find this to be particularly problematic since they believe that these journals may be what are often called "predatory journals", and therefore may threaten the credibility of science, by polluting the database with "weak research". Therefore, the authors aimed to compare the number of citations per year, and per journal, and per document, before and after the journal was delisted. They conclude that the number of citations was actually higher after the journal was discontinued from Scopus. Although we understand the problem these authors try to highlight, we have some major concerns regarding some aspects of this study (mainly related to baseline assumptions and lack of clear definition) and also some minor ones.

We are very grateful for the insights of the reviewers which have led us to improve our manuscript. We have now submitted a revised version of the manuscript and herein is our point-by-point reply to the comments. English form was also revised.

Major comments:

Baseline assumption: In this article, the authors suggest that if a journal is being indexed, even for a long period of time (half of them have been indexed for 8 to 54 years), and is encountering "publication concern", then all the previously published articles should become suspicious of bad science. We are not sure this should be considered straight forward, for the reason developed under our second major concern.

Reply: Thank you for the opportunity to clarify this. We made no such claim in the paper and, in fact, our key message was different. We clearly stated in the discussion that "It would be unfair to punish researchers for an eventual deterioration in journal performance; changes in the standards employed by the journal may change over time and the researchers may be unaware of quality issues". We aimed to provide an analysis and describe the main scientific features and citation metrics of journals discontinued from Scopus for publication concerns as we strongly believe that this phenomenon merits discussion. We fully agree with the reviewer that further evaluation is required before the best solution for all aspects of this complex issue is determined. In fact, this study is the first to provide some of the information required to answer this question, albeit not all. We also agree with the reviewer regardless of the solution that is decided upon in the future, it should ensure that researchers are not unfairly punished. This is clearly stated. However, as we also point out, this issue can no longer be ignored; it involves a large number of journals and published documents.

predatory journal is, remains. The authors use different sources to try to identify journals as "predatory" and we can only realise that the sources do not seem to agree. Although authors auto cite their own "consensus definition" of predatory journals and publishers "(..) entities that prioritize self-interest at the expense of scholarship and are characterized by false or misleading information, deviation from best editorial and publication practices, a lack of transparency, and/or the use of aggressive and indiscriminate solicitation practices." they fail to underline that not everybody agrees with this definition. Also, the recent COVID-19 debacle of very low-quality scientific publications, published in usually highly regarded journals, suggests that bad peerreview and misleading articles may not be a characteristic of any journal.

Reply: First, we would like to highlight that the definition we reported for "predatory" is not the authors' own. It was taken from an international collaboration of 35 authors who extensively studied the topic. Although we agree that no definition is perfect, this is most certainly not something we decided on ourselves and a consensus process was involved in its determination. If the reviewer wishes to argue with the definition provided, this should ideally be taken up with those involved in the consensus process. We surveyed recognized lists (i.e. Cabell, updated Beall, DOAJ) to evaluate the extent of predatory journals among the discontinued journals. With regards to the comment regarding the quality of COVID-19 research: Very true. We too have been following this topic with great interest. However, two wrongs do not make a right. In fact, this precise issue makes the discussion of journal metrics and our responsibilities towards them even more pertinent. Our research highlights some of the issues that arose with monitoring of the publication process from a different angle. It also promotes the need to continue to increase awareness within the scientific community itself regarding the damage that could potentially be caused by low-quality papers.

Also, it remains unclear to us how a journal may be indexed for 8 to 10 years, and all of a sudden become "predatory". Or was it predatory in the first place, but was only uncovered after such a long time? If this is what the authors suggest, then what should we think about "recently" indexed journals? They may all be predatory as well, and will only be uncovered in 5 to 10 years? Reply: It is our impression that the process may occur in two manners: (1) Some of the more recently indexed journals may indeed turn out to be predatory. So indeed perhaps newly indexed journals need to undergo more rigorous monitoring than well established journals. Whether our impression is correct and, if so, how this should be done, are questions far beyond the scope of our research; (2) Some of the discontinued older journals probably did deteriorate slowly. Our impression was that this process is typically a "slippery slope" and does not have an abrupt cutoff. As our analysis was not intended to study this question, we prefer not to speculate on the ideal timing for journal discontinuation. More data and expert input is needed on how to identify this process in the future.

Comment 4: "Publication concerns": This term needs to be better defined in order to really understand what lies behind it. It remains unclear why these journals have been excluded from the Scopus database at some point. Interestingly, half of these journals have been deemed good enough to figure in the database for more than 8 years… that's a lot! And all of a sudden, they are not judged acceptable anymore and are discontinued from Scopus. Ok, why not. It may take some time before someone alerts Scopus of the misbehaviour of a given journal, although more than 10 years for 25% of them seems a lot. Or could it be a problem behind the vague concept of "publication concerns"? Could it be that the publication has stopped? Or the journal has changed its name? Or has merged with another one? Or has changed in quality over time? Illustrating some of the reason for discontinuation would help the reader understand the context. Reply: The term 'publication concerns' is not one which spontaneously decided upon. It is the term defined and used by Scopus. Indeed, we report in the manuscript all the available definitions and details provided by Scopus. Unfortunately, no additional details are publicly available regarding the criteria used to discontinue a journal because of 'publication concerns'. We too would be delighted to receive more details as they may be important.

Having said this, we honestly doubt that merging with another paper or changing a journal name is cause for publication concern. With regards to the reviewers' rumination on the time gap for discontinuation: As noted above, it is indeed possible that some journals have changed quality over time or that they were evaluated only several years after indexing. This information would most certainly be interesting if it were publicly available, but it is not. Furthermore, as also stated above, this is not within the scope of our project. Comment 6: Increase in citations: The authors are worried that the citations of these journal have increased after the journal's indexation was discontinued in the Scopus database. The problem here is that they do not seems to consider the fact that this may be the case for all journals (those indexed and those discontinued) which is probably due to the rapid increase in the number of publications over time. Unfortunately, this study lacks a "control group" (journals whose coverage has NOT been discontinued in the Scopus database) which could have help the reader understand whether the increase in citation of these journals was similar, was higher, or was lower than that of "legitimate journals". Reply: Please see below our response to this and the next comment together.

Comment 7: Underlying discourse: The term "inflated" used in the title, in Figure 2 and conclusion suggests manipulation or distortion of citations and an artificial advantage for authors of articles published in predatory journals before they are removed from Scopus. This is not demonstrated by the reasoning and data used in the article as a basis for comparison is missing.

Reply: Indeed, the lack of a control group is a study limitation. We now point this out in the discussion section (see page 13). However, the authors have no interest vested in presenting an "underlying discourse" we have taken this comment very seriously. We have now removed the term "inflated" from both the title and the conclusions. We also modified the caption of Figure 2 , substituting 'can' with 'may'. Our decision to submit the full database for publication and to select an Open Research publishing platform stems from precisely this reason -we would be delighted if this study was repeated and expanded on in the future. We calculated "citations per year" as the ratio between the total number of citations (before discontinuation plus after discontinuation) and the number of Scopus years. In the revised version of underlying data table 1 we have now added a box with a more detailed description to enable the readers to repeat our analysis. However, we must point out that online data changes daily. Therefore, in order to reproduce the data to perfection, one would need to know for which one of the 317 journals that we studied -on which day through the duration of the study period we downloaded the data. The overall process took about a month as described in the paper. This issue may render the data not reproducible to the dot. However, at any time of examination, the overall trends should remain the same.

Comment 8.2: Some variable names and analysis are misleading, such as "Actual Pubmed", described in methods section as "inclusion in PubMed" and in table 6 as "main database indexing". It does not reflect whether the journal is currently indexed in PubMed, but may in some cases only indicate that a single article is present in PubMed or selected citations, due to their deposit in PMC (eg. "Advanced Materials Letters"). Reply: Thank you for pointing out this omission. We have now revised both the manuscript and underlying data table 1 to specify that we collected data on the inclusion of articles in PubMed. We also changed the title of Table 6 as follows: " Table 6 . Discontinued journals' current Open Access policy and the indexing of their articles in major databases". erroneously written as '43451' rather than '4345'. This led to the number of 170 citations per year for the period before discontinuation. We have corrected the resultant calculations. We have also re-checked the database for additional typos (none were found). The main findings of the manuscript did not change after this correction. Nonetheless thank you for pointing out the mistake.

Regarding Advanced Material Research -we have re-checked the data and confirm that it is correct. We do not have an explanation for the huge difference between the period preceding and succeeding discontinuation.

Minor comments: Comment 4: Methods: The use of the term "discontinued" both for DOAJ (Results) and for journal publication (Background) is confusing. Should we say "excluded" or "delisted"? Reply: The term 'discontinued' is that used in the Scopus database. The label 'coverage discontinued in Scopus' is also displayed on the discontinued journals' page. The downloadable list of journals whose coverage has been discontinued is also named by Scopus as 'Discontinued sources from Scopus'.

As it is important that the labels used in the manuscript remain consistent with official labels and definitions, we felt we could not change the term 'discontinued'. We did not give a subjective definition of 'document' but included all the indexed documents provided by the Scopus database. We have added this detailed description in the methods section of the paper, and we also specified the calculations performed in underlying table 1.

Conclusions: it's unclear how the conclusion regarding "predatory journals" is drawn. Also, we don't think the career advancement are "artificial", they are real! Although maybe "undue"? Reply: As a result of this comment we have modified the conclusions to state as follows: "Journals whose coverage in Scopus has been halted for publication concerns continue to be cited. This paradox may influence scholar metrics, potentially prompting career advancements and promotions. Further studies are needed, also investigating the journals discontinued from Scopus using the criteria "outlier performance -radar", particularly effective in flagging potential predatory journals. Countermeasures should be taken to ensure the validity and reliability of Scopus metrics for both journals and authors due to their importance for scientific assessment of scholarly publishing. Creative thinking is required to resolve this issue without punishing authors who have inadvertently published good quality papers in a failing or predatory discontinued journal." Comment 8: Introduction: "publications from no longer indexed journals may not be removed retrospectively … hence articles … could remain part of the database7" p.3 -> this conditional statement seems to contradict abstract which categorically states "These journals remains indexed" as well as the author's conclusion "we propose that CSAB could apply these measure case-by-case". Reference 7, linked with statement, was not helpful to clarify. Reply: Thank you for pointing out that the language in this sentence requires improvement.

We have revised this to read more succinctly: "The list of the discontinued sources is publicly available and is updated approximately every six months 6 . However, articles published in journals that were discontinued and are no longer indexed, are probably not removed from the Scopus database. It has been claimed…" Comment 9: Methods: "Independently collected by eight of the authors in pairs": not very clear: two by two, or checked by two different people independently? Reply: Agree. We specified that four pairs of authors independently collected the data (i.e. two people independently collected the same quarter of the data. The entire database is the result of eight people collecting the data).

Comment 10: "the year of our data collection": more precision maybe? Reply: Agree. We changed "the year of our data collection" with "2020".

Comment 11: Results: Why were data from 31 journals not retrieved? What was the problem?

Reply: The journals were not found on Scopus database using the search tool. This is now also stated in the paper. The relation between the publisher and de-indexing of articles in Scopus after discontinuation is an important question that should be addressed in further research. We aimed to provide a snapshot of the effect of ongoing article availability, rather than explore publisher and/or Scopus policies associated with journal discontinuation. Table 3 : don't need 2 decimal precision in %.

Reply: This has been changed in accordance with the reviewer's request. Reply: Again-we are grateful for the reviewers' sharp eye. We rechecked the data and found a typo: The number of citations after discontinuation was reported as 607621 when it should have been 607261 (this can be seen in our underlying data table 1 and in the main text). We have corrected this in the new version of the manuscript.

Citations by year in table 5: The number of years before and after the journal is removed from Scopus is very different, the average is more than 9 years before and 4 years after (median of 8 and 4 respectively) which makes the comparison in Table 5 not relevant. Indeed, the number of citations per year is higher during the 2 or 3 years following the publication of the article and decreases sharply with time (DOI:10.1371/journal.pone.01537302) so that the ratio of citations per year also decreases if a larger number of years is used.

Reply: This is probably true. Although some papers may undergo resurgence this is probably not common. However, we found no better way of coping with the issue of the different number of Scopus coverage years across journals. And furthermore, this "decay" is likely to be fairly consistent across all journals both before and after discontinuation. As we also added to the study limitations the lack of a control group of non-discontinued journals, we have also inserted a word of caution regarding our results.

Comment 19: Distribution of articles: of the 317 journals analysed, 5 contain more than half of the articles concerned by this question. This very inhomogeneous distribution means that the statistical analyses and the percentages per journal do not take this kind of distribution into account. Reply: We used non-parametric tests to reduce the effect of non normal distribution of data on our findings precisely for this reason. We also present IQRs and ranges to be more informative.

claiming"? to our understanding the article does not say that open access systematically means predatory. According to ref 9 et 22, the large majority of DOAJ indexed journals were not found in Beall's list or Cabell's Blacklist. Reply: We have changed the phrase "Of greatest concern is our finding that many of the discontinued journals display predatory behaviours in claiming to be open access" to "Of greatest concern is our finding that many of the discontinued journals display predatory behaviours in claiming to be open access without actually being indexed in DOAJ."

Comment 25: p. 8: "Such journals" unclear: predatory journals or OA journals? Reply: 'Predatory'. This is now written.

The authors highlight that a limitation of their methodology is that they have included the year of discontinuation in the period "after discontinuation", which could have led to overestimations. Then why not present the 2 analyses with the year of discontinuation included in the period BEFORE and in the period AFTER discontinuation, so that the reader can check for himherself what bias this has induced? Reply: We mentioned the possibility of overestimation in order to be entirely honest. However, our impression, just from eyeballing the data during collection, was that this would not lead to much of a change. More importantly, in the early stage, we planned no such analysis and therefore did not collect the data that would be required to do this analysis. At this stage performing such an analysis practically requires that the data be recollected in near entirety again which is no simple task.

Reply: This question again is one of policy and therefore not in the scope of our study. The reviewers' comments indeed present much food for thought in terms of future research. Solutions should also address metrics and citations deriving from these journals. We have therefore added that while this may be an immediately implementable temporising measure, additional thought should be dedicated to address of these aspects as well while maintaining fairness. 

How Scopus works

Dimensions: A competitor to Scopus and the Web of Science?

From Excessive Journal Self-Cites to Citation Stacking: Analysis of Journal Self-Citation Kinetics in Search for Journals, Which Boost Their Scientometric Indicators

How Frequently Are Articles in Predatory Open

We would like to thank Dr. Antonio Corrado ("Korrado 20") for creating and providing the Table 6 : maybe a good idea to separate the "positive" facts (being indexed in Pubmed, WoS or Cabell's whitelist) from the negative ones (Beall's list, Cabell's blacklist DOAJ discontinued.) Reply: This was a good idea. We have changed the order of the items in Table 6 accordingly.

Discussion: p.7 Unclear why the term "ghost journal" suddenly appear and how it is defined. Reply: We modified the phrase; the text now reads "The fact that discontinued journals might help to move up in academia is a relevant issue, and has inspired the vignette depicted in Figure 2 : "Discontinued journals may inflate authors' metrics lifting them unnaturally and effortlessly." We decide not to remove the term 'inflate' from Figure 2 caption as the figure is an allegorical and so intentionally exaggerated representation of the phenomenon.Comment 24: p. 7-8 "Of greatest concern is our finding that many of the discontinued journals display predatory behaviors in claiming to be open access" -> do you mean "displaying… are

Reply: Bizarre or not, the data is what it is. Here we describe how the data was collected in detail:We searched the name of each journal using the Scopus database "Sources" page ( https://www.scopus.com/sources.uri?zone=TopNavBar&origin=searchbasic), then we opened the page of the journal and selected "view all documents".

At the following page, we checked the box "All" to select all the documents and then clicked "View citation overview".

At this stage, two options may appear: a) a page with a line chart reporting the number of citations per year, or b) a flag inviting a request for a citation overview download due to the large size of the overview which does not allow on-site display as a line chart.

In the a) situation, we manually summed the citations displayed. In the b) situation, we inserted in the request form an email address and Scopus sent us back a .csv file for each request, containing the requested data.

As this detailed description add little to the main text of the paper we have not added it to the description of the methods. We wish to reiterate the point made in our previous response -that data collection was not completed in a single day. The data for this study required accessing several sources for each paper, documentation of different variables from each source and performance of individual calculations. For this reason, we report the time frame during which the data was collected (24th January -end of February 2020). Citations are continuously updated as they grow and new journals are indexed in the database, so at any time other than the exact time we downloaded the data, the number of citations would be slightly different than ours. Specifically, with regards to the examples given for "bizarre" data, we have re-checked these items and our replies are as follows: Our table 1 did not and does not state "80 citations per document before discontinuation" for Mental Health in Family Medicine. We report a total of 175 documents, with 213 citations before discontinuation and 14123 after discontinuation. Thus, 1.22 citations per document before discontinuation and 80 after discontinuation.

Regarding Pharmacognosy Reviews -We are grateful to the reviewer for picking up on this typo -In the table, the number of citations before discontinuation was 2. 

This manuscript has a long list of authors, many of which have published (1) together (2) on this topic. Hence the probability of citing previous research by these authors is high. Is the reviewer suggesting that researchers never base their newest work on their previous work? Experts that consistently research any field of research are likely to work together and to cite their own work -each step supports the next.

Despite the nature of this comment and our concern that it may detract from the quality of our paper, we have removed some of the references. The authors of this paper have no need of self citations -for promotion or any other purpose. This is a very relevant study in the growing scholarship around predatory publishing. It is one of the first studies that look at how predatory or at least questionable journals that have been delisted from citation database continue to have a presence in academia. More specifically, the paper asks the very important question why databases like Scopus (and others) continue to track the citations of journals that have been removed. This creates a distortion at many levels, including at the individual publication profile level.I am happy to recommend indexing of this paper as it is (some minor grammatical editing is required).

Are all the source data underlying the results available to ensure full reproducibility?

Competing Interests: No competing interests were disclosed. Reply: We thank the reviewer for the comment. We have now edited the manuscript and corrected errors and typos.

The benefits of publishing with F1000Research:Your article is published within days, with no editorial bias • You can publish traditional articles, null/negative results, case reports, data notes and more • The peer review process is transparent and collaborative • Your article is indexed in PubMed after passing peer review • Dedicated customer support at every stage • For pre-submission enquiries, contact research@f1000.com