A journal that adheres to these recommendations may indicate so in its instructions to authors. In fact, both URM [1] and CONSORT [4] encourage adhering journals to indicate so in the instructions to authors, and COPE [3] suggests that journals use the instructions to inform authors about specific editorial policies. The instructions to authors and related editorial policy statements are documents in which journals typically present themselves and provide information about formatting a manuscript [9]. These are also documents to which authors can refer for guidance on the scientific content of papers. However, the amount of guidance provided varies greatly [9]. Well formulated instructions distinguish a journal for professionalism and rigor and may be considered as evidence of “editorial leadership”. Whether a high level of editorial leadership results in a better quality journal is unknown.

To investigate a possible relationship between journal quality and editorial leadership–as demonstrated in the instructions to authors, biomedical research journals from the UK and Italy were assessed and compared. These two European nations are similar in population and gross domestic product [10]. Both are members of the Group of Eight Industrialized Nations and are considered “scientifically advanced” [11]. Both also have legislation regarding ethical practices in biomedical research, as established by European directives [12]. However, the countries differ in language, culture and other characteristics relevant to research and publishing. For example, many initiatives to promote quality reporting, such as the EQUATOR network [8] and COPE, have roots in the UK. The UK spends almost twice as much as Italy on research and development [13], [14]. In Italy, underfunded research and low use of meritocracy [15] lead to “brain drain” (i.e. emigration of the best researchers to countries with better research environments), a phenomenon characteristic of Italian science [16]. Italy has no leading general medicine journal, like the BMJ and Lancet in the UK. Finally, unlike the UK, Italy has no member journal in the ICMJE and no affiliate in the World Medical Association, which produces the Declaration of Helsinki. These differences could negatively influence journal production in Italy and might be reflected in a lack of editorial leadership.

The two groups were similar in terms of the numbers of journals publishing research involving both humans and animals (biomedical), involving only humans (clinical), or not involving humans (e.g. animals or plants) (Table 2). However, Italian journals published fewer articles annually (median, 60 vs. 93; p<0.006) and were older (p<0.001) than UK journals. In particular, 8 Italian journals had been started since 2000 compared to 20 UK journals. These results suggest different trends in innovation and turnover. Almost all UK journals had an international editorial board and, although 47 Italian journals also did, this difference was significant (p<0.001). Most articles published in UK journals had a non-national first author, whereas only 37% of examined articles in Italian journals were authored internationally (p<0.001). These results suggest that UK journals can be considered international, while Italian journals have a tendency to being international, as also evidenced by their preference to publish exclusively in English. Concerning use of the Internet, Italian journals less frequently had online archives or links from PubMed to the archives. In particular, 17 of 43 Italian journals with online archives lacked PubMed links, vs. 6 of 74 UK journals. Despite being older, Italian journals offered archives covering fewer years (median, 6 vs. 10; p<0.001). However, Italian journals were more likely to offer open access and less likely to impose an embargo period, a phenomenon almost exclusive to UK journals. Regarding type of articles, journals from the two countries published similar, low percentages of letters to the editor (p=0.070), but in Italian journals clinical trials were less frequently randomized and controlled (p=0.029). Regarding citation, fewer Italian journals were indexed for IF (28 vs. 54, p<0.001) and those indexed had a lower median IF (1.2 vs. 2.7, p<0.001). All UK journals and 75 Italian journals were indexed in the SCImago database. Italian journals scored significantly lower in SCImago journal rank, H index and number of citations per document (p<0.001). Together, these results show that Italian journals are “smaller” and score lower for quality than UK journals.

Each journal's instructions to authors and editorial policy statements were examined to assess the editorial leadership demonstrated towards authors (Table 3). Italian journals were significantly more likely than UK journals to declare to adhere to URM (27 vs. 11, p=0.003). However, 24 Italian journals cited an outdated version (1997 or earlier) and 3 provided no reference; only 3 linked to the ICMJE website. UK journals also failed to cite the current version (5, no citation; 6, outdated versions) but 7 provided links. Fewer Italian journals based authorship on a “substantial contribution” or “scientific responsibility” (p=0.019) or required that manuscripts specify authors' contributions (p=0.005), funding or sponsorship (p<0.001), and competing or conflicting interests (p<0.001). For journals publishing human research, similar low numbers required that manuscripts state that research adhered to the Declaration of Helsinki: 9 Italian journals referred to outdated versions (1983 or earlier) and only 1 Italian and 6 UK journals referred or linked to the current version. Fewer Italian journals inquired about ethics committee review (p<0.001). Some inquiry about informed consent was made by 16 Italian and 42 UK journals (p<0.001). Of these, 7 Italian and 20 UK journals required informed consent for the publication of personal data, and 7 and 9 journals, respectively, required informed consent for patients' participation in clinical trials. No Italian journal required registration of clinical trials vs. 20 UK journals (p<0.001). For journals publishing animal research, fewer Italian journals inquired about adherence to animal research laws (p=0.082). No Italian journal adopts CONSORT or QUOROM, while 15 UK journals follow CONSORT and 11 QUOROM. No Italian journal but 33 UK journals adhere to the guidelines of COPE. These results document that Italian journals show less editorial leadership than their UK counterparts, but that UK journals have room for improvement.

To further examine the relationship between editorial leadership and quality and to understand the impact of potential confounders inherent to a cross-cultural comparison, multiple regression analysis was performed for the 73 Italian and 69 UK journals that published research on humans (Table 4). When SCImago journal rank was used to define quality, the 12 parameters of editorial leadership considered as predictors explained about 37% of the variance in quality among the journals (multiple R=0.609, p<0.0001). Each potential confounding variable added singly to the model had a minimal effect, and when all 5 confounders were added the ΔR² was only 6.2%. With SCImago “cites/doc” as the definition of journal quality, editorial leadership explained almost 50% of the variance; again, the possible confounding factors had a small effect and together increased R² by 13.8%. Finally, for the subset of journals indexed for IF, editorial leadership explained 49.9% of the variance in this parameter and confounders increased this to 57.4%. These results suggest that editorial leadership is intimately associated with journal quality as assessed by various citation statistics, and that the impact of variables such as country of origin, publishing language and internationality is not strong.

A unique feature of this study is its focus on research journals produced in two countries. Previous studies on editorial practices and quality focused instead on leading journals (i.e. selected subjectively or by IF [9], [18]) or on randomly sampled journals [19], but lacked comparison groups. The comparison between two European countries provides insight into the editorial challenges faced by journals from non-Anglophone industrialized countries. Furthermore, since the selection criterion in this study was indexing in Medline, rather than IF, the samples are larger and better representative of the quality and diversity of biomedical research journals from the two countries. Finally, only journals both edited and published in the countries of interest were considered. This dual requirement excluded 58% of Medline-indexed journals listed as being from the UK and 8% of journals listed as being from Italy. Thus, when characterizing journals (and national research output), especially from the UK, it is important to verify country data and not rely on information provided by sources that only use the publisher's country of origin.

While the findings of lower quality and editorial leadership among Italian journals may have been expected [20], [21], the outcome for UK journals was surprising. In particular, few UK journals (14%) declared to adhere to URM. This rate is significantly lower than that for Italian journals (36%) and is also lower than the 41% reported by Schriger et al. [9] in a survey of instructions to authors from 166 leading general and specialty clinical journals. Moreover, just less than half of UK journals (like Italian journals) requested manuscripts to indicate that research adhered to the Declaration of Helsinki, despite the fact that most journals in both groups publish research on humans. Notwithstanding these low rates, UK journals often adhere to specific items in both URM and the Declaration of Helsinki, such as requiring manuscripts to indicate sources of funding (72%), demonstrate approval by ethics committees (66%), disclose competing interests (63%), and document informed consent (55%). The finding that 20% of randomly sampled UK journals endorses the CONSORT statement agrees with the 22% observed by Schriger et al. [9]. That 63% of UK journals had a written policy on the disclosure of conflict of interest is a positive finding, higher than the 33% found for 84 high-IF journals from 12 scientific disciplines [18]. Why few UK journals–and even fewer Italian journals–provide authors with adequate guidance on the scientific content of papers is unknown: they may trust authors and peer reviewers to handle the issues, they may ignore the issues entirely, or they may believe that setting high standards will reduce the number of submissions [9]. This was the case for one editor from a developing country, who feared that adhering to the requirement for clinical trial registration would result in authors sending their papers to journals with lower standards [22].

This study is limited by its observational design, which is unable to demonstrate a causal relationship between editorial leadership and journal quality. Another limitation is that the study only evaluated leadership as expressed through the instructions for authors and did not investigate actual editorial practices, such as systems of in-house review, peer review and technical copyediting; thus, it may have underestimated leadership in journals with “author-helpful” policies [23]. Additionally, this research used publicly available information, rather than interviews or questionnaires. Judging a journal by its written policies may not completely describe its editorial practices, as highlighted by the study of 84 high-IF science journals, of which 11 had only unpublished conflict of interest policies [18]. Furthermore, journals do not always adhere to their written policies, as shown by a survey of 5 leading general medicine journals: despite adequate instructions, 31% of published papers did not mention ethics committee approval and 47% did not mention informed consent [24]. Finally, only 20% of the total estimated number of UK journals was evaluated; thus this study provides a full survey of Italian journal production only.

Nation-wide surveys of journal production are rarely performed. Instead, the scientific output of nations is evaluated from the numbers of published papers and their citations [13], [25] or from the journals in which papers are published (e.g. [26], [27]). These evaluations emphasize publishing in high-IF journals, nearly all of which are based in Anglophone countries. Although such journals are considered “international”, they have limited coverage of health issues relevant to the developing world [28]. This criticism has encouraged a new appreciation for the real and potential contribution of “local” journals–especially in resource-poor countries–and has led to the creation of associations for editors in Africa and the Eastern Mediterranean [29]. In this panorama, journals from non-Anglophone industrialized countries (especially in Europe) have an unclear status. In a recent debate, journal production in Germany was given the status of a developing country, especially for difficulties in scientific communication [30]. Yet, as shown here, these journals can have a substantial international character and aim to an international audience. They may act as a bridge between mainstream science and the scientific periphery [31], by publishing papers from less rich countries, especially through editorial collaborations with Eastern Mediterranean and Asian medical associations [32], [33]. Although this author community may produce lower quality research (due to poor resources or research skills), the information can be internationally important [34]. Thus, it is imperative that these journals, like those from semi-developed nations [35], adopt good editorial practices, to improve their scientific quality as well as their effectiveness as an international voice for non-Anglophone authors. Assistance in reaching these goals is now available from the EQUATOR network [8].

Appendix S1