Most-cited public health articles of scientific journals from Brazil

ABSTRACT OBJECTIVE To describe the most-cited articles in public health scientific journals edited in Brazil. METHODS Articles published between 2008 and 2010 by public health journals edited in Brazil and indexed in the Scopus database were included, and citations received up to five years after publication were ranked. We studied a total of 105 articles, as the last seven articles shared the same number of citations and so were given the same rank. RESULTS The most-cited articles received a median of 28 citations, and the distribution ranged from 22 to 95 citations. These articles describe advances in the areas of Epidemiology (74%), Health Policies, Planning and Administration (19%), and Social and Human Sciences in Health (7%). Only half mentioned that they have received funding. About 75% of the articles were written by three or more authors and 90%, by authors affiliated to public institutions such as universities and government organizations. Fifteen individuals were responsible for authoring or coauthoring three or more of the 105 articles studied. The journals Cadernos de Saúde Pública, Revista de Saúde Pública, and Ciência & Saúde Coletiva have published the vast majority of the most-cited articles in the area (94%). CONCLUSIONS In Brazil, the most-cited articles in public health mainly report Epidemiology research, are written by groups of authors and by researchers affiliated to public institutions and are published in journals with a greater impact. Periodical analyses of these data can show potential changes in the characteristics of articles that most attract public health scientists.


INTRODUCTION
Citations are considered an indirect measurement of the contribution of an article to the knowledge generated in the field, i.e., the link between the finding of the investigation and its importance for science 1 . Therefore, citations have been widely used and valued to provide indices and impact indicators in science 2 . Garfield and Welljams-Dorof 3 have identified a significant correlation between the quality and relevance of an article and the number of citations it receives.
By analyzing articles that exert the greatest influence on a field of knowledge, we identified where, how, and by which scientists this field moves forward. Thus, several studies have evaluated highly-cited articles in specific fields such as Rheumatology 4 , Gastroenterology 5 , Psychiatry 6 , Surgery 7 , among others. The most-cited articles represent a high degree of research impact and an effective investment of resources 8 .
Public health is a field of production of multi-and interdisciplinary scientific knowledge 9 , based on epidemiology, social and human sciences in health, and policies, planning and administration in health 10 . This field provides an excellent opportunity to evaluate citation performance among areas. Smith and Leggat 11 evaluated the 10 most-cited articles in public health, but the authors included articles from only one journal.
In Brazil, few studies have addressed the question of citations, and they are usually from the areas of Information Science, Bibliometrics, and Scientometrics [12][13][14] . In specific fields such as public health, some authors, such as Coimbra Jr 15 , Barata 16 , Iriart et al. 10 , Cuenca et al. 17 , have been analyzing the citations attributed to journals and articles, but few of them address the most-cited ones.
This study aimed to characterize the most-cited articles published in public health journals edited in Brazil. Our findings should help students, young researchers, and readers in general to become familiar with the authors, institutions, and journals that have made a leading contribution to public health in Brazil. Also, this study may provide researchers, research funding agencies, and university administrators with insights into the research trends in this field.

METHODS
Following the trend of bibliometric studies considering the so-called "top 100" of an area of knowledge, we present here the most-cited articles published in Brazilian public health journals.
To assess those articles, we selected the public health scientific journals that are edited in Brazil, according to the Associação Brasileira de Saúde Coletiva (Abrasco -Brazilian Association of Collective Health), which have been indexed in the Scopus database since 2008. This database was chosen because, besides being one of the main sources of bibliometric data, it indexes more journals edited in Brazil. The initial year was chosen to increase the number of journals represented since fewer journals were edited in Brazil on this database until 2007. Thus, the following journals were selected: Cadernos de Saúde Pública; Ciência & Saúde Coletiva; História, Ciências, Saúde -Manguinhos; Interface: Communicação, Saúde, Educação; Physis; Revista Brasileira de Epidemiologia; Revista de Saúde Pública; and Saúde & Sociedade.
For the retrieval of the articles, we searched for records classified by Scopus as original articles or review articles, published between 2008 and 2010. The final year was considered to include citations received up to five years after publication since citations usually do not occur soon after publication. So, for articles published in 2008, we studied citations received until 2013, and so on. This standardization provides, for all articles, the same time availability to receive citations. Initially, we retrieved 3,242 records and, after the age adjustment, we filtered those articles in descending order of citations and selected the 100 most-cited ones. However, seven papers ranked last in the classification, as they contained the same number of citations (22 citations). Overall, in this study, we included the 105 most cited articles.
We performed the descriptive analysis of the articles included according to the following variables: citation ranking, publication year, publication language, publication type, thematic category, journal title, number of authors (one, two, three, or more), authors' names, and funding. The type of publication was classified into two stages: first, we evaluated whether the article was methodological or not; second, if it was not methodological, we evaluated whether it was an original study or a review. The thematic addressed in the articles was based on the categories defined by Abrasco for the area of Collective Health, namely: epidemiology; policies, planning, and administration in health; and social and human sciences in health. The first author's affiliation was classified as a public university, private university, hospital, governmental institution, or non-governmental institution. The funding institutions were categorized into: development agency, such as Research Support Foundations (FAP) and Brazilian National Council for Scientific and Technological Development (CNPq); national governing bodies, such as ministries and state departments; foreign or national institutions, such as World Health Organization; and scientific associations or societies.

RESULTS
Among the 3,242 articles published between 2008 and 2010 by the journals included in this study, the most-cited articles (n = 105) received between 22 and 95 citations, with a median of 28 (p25 = 24 and p75 = 36). The 105 most-cited articles are listed in the Table 1.
As described in Table 2, of the 105 articles, most were original articles (70%), in the area of Epidemiology (74%), written by three or more authors (74%) and published in English (57%). All have abstracts in Portuguese and English, and 34 (32%) also have an Abstract in Spanish. It is worth noting that, despite standardizing the citation window (citations up to five years after publication), we identified a few of the most-cited articles published in 2010.
Cadernos de Saúde Pública, Revista de Saúde Pública, and Ciência & Saúde Coletiva have published the vast majority of the most-cited articles in the area (94%). Another group is constituted by the Revista Brasileira de Epidemiologia, Interface, História, Ciência & Saúde, Physis, and Saúde e Sociedade, the articles in which seldom if ever feature among the most-cited ones. The distribution of the most-cited articles according to the journal is shown in Table 3.
Regarding the affiliations of the first author of each article (Table 4), public universities represented 67% of the institutions. Authors were also affiliated to governmental organizations (23%) -such as the Fiocruz Foundation, the Ministry of Health and health departments -and to private undergraduate institutions (6%).
Regarding research funding, 52 articles reported some kind of funding; of these, half reported support from two or more institutions. The most frequently mentioned funding agencies or institutions were FAP (35/52) and CNPq (31/52). Support from foreign agencies was mentioned in 25 of the 52 articles (48%). Brazilian government agencies accounted for 21% (11/52) of these grants, and a research incentive association accounted for 1 grant (data not shown in table).
In relation to authorship and coauthorship, in total, 388 individuals were responsible for authoring or coauthoring these articles. Table 5 shows the most-productive researchers (n = 15). In addition, eight other researchers were responsible for 1 authorship and 1 coauthorship; 26, for two coauthorships; 84, for 1 authorship; and 255, for 1 coauthorship (data not shown in table).

DISCUSSION
The number of citations received by public health articles differs significantly from that found by studies in diverse subjects and countries, varying from about four thousand citations received 5 to one thousand 18 , from the medical field, but over an extended period, stretching back to the 1950s. In a bibliometric study of systematic reviews, Royle et al. 19 found an average of 26.5 citations after four years of publication. The authors used the same information source, the Scopus database, and a similar analysis period (year of 2008) to that of our study. In our study, median citations were equal to 28. Although we included a period longer than that of Royle et al. 19 , which could overestimate the total citations of the articles of our study compared with the one of Royle et al. 19 , this overestimation would be counterbalanced since systematic reviews tend to receive more citations 20,21 . In fact, the number of citations potentially reflects the process of creation and dissemination, which differs between the areas of knowledge 22 , the periods covered and the data sources, among other factors.
Less than 5% of the most-cited articles have only one author, while most (74%) have three or more authors. This result corroborates the literature on the positive influence of the number of authors per article on the number of citations received 23 . Regarding language, more than half were published in English (part of them, bilingual). Clearly, it denotes the effort for internationalization by the journals edited in Brazil.
The proportion of the most-cited articles reduced from 2008 to 2010. Certainly, the more recent, the less time the article would be found, read and quoted more often and by more authors. However, for all articles, we counted citations received up to five years after publication, reducing the bias of publication year. Thus, this result may have been due to the number of article journals published during the years considered in this study. In fact, the two journals that contributed the most to the list of the most-cited articles published a total of 605 articles in 2008, 521 in 2009, and 422 in 2010; that is, the number of most-cited articles decreased as the total of articles also decreased.
As for the inf luence of the impact factor of journals on the citations, there are controversies. According to Callaham et al. 24 , the greater the impact of the journal, the more its articles are cited. The authors reported the impact factor is more influential even than variables related to the quality of the investigation, such as the study design and other methodological aspects. As a matter of fact, the two journals with the highest  18 , when evaluating the most-cited articles in the medical field, also found most of these were published by journals with a high impact factor. For Larivière and Gingras 25 , this influence can result both from the better dissemination of the journals that have a greater impact, and from the citing researchers' perception that the articles published in these journals have higher quality.
Slyder et al. 2 , however, emphasize that being published by high impact journals does not guarantee that the article is highly cited since the citation rate varies greatly between articles. This indicates that other features in these articles make them attract more citations.
The study by Royle et al. 19 corroborates this information by showing that, although impact metrics have explained more than half of the variation in citations, they do not accurately represent the number of citations of individual articles.
It should be noted that, when analyzing the journals in which the most-cited articles were published, a consistency can be seen between the order of the most-cited articles and the number of those that did not receive any citations 17 , i.e., the journals that published more articles with the highest number of citations were also the ones that published fewer articles that have never been cited.
Regarding the sub-areas, Iriart et al. 10 , in a recent evaluation of the scientific production of postgraduate programs in the public health area, found a similar tendency in the proportion of researchers in each discipline: 49% belong to Epidemiology; 20%, to Health Policies, Planning and Management; and 17%, to the Social and Human Sciences in Health. This proportion is reflected, in this study, by the distribution of the most-cited articles according to the sub-areas.
The search for institutional excellence has aroused interest in identifying the institutions performing best in the development of science in a country, or in a specific area. In public health in Brazil, public universities stand out as science generators. Packer and Meneghini 12 , in 2006, had already identified the influence of public universities on the impact of articles. Both in this paper and in the study by Packer and Meneghini 12 , the presence of private higher education institutions in the most-cited articles was also observed, although to a lesser extent.
In Brazil, public health research is subsidized by development agencies, national government agencies and mainly by public resources 26,27 , as corroborated by our study. We can infer that this is a characteristic of developing countries, such as Brazil, and of a multidisciplinary area, such as public health, while private financing is correlated with the highest citation index in the area of cardiology in developed countries 28 . However, some journals do not adopt the requirement of funding acknowledgments, and this fact may have led to the large proportion of articles classified as being without financial support 27 .
As for the type of article among the most-cited, we selected a priori the categories original articles and reviews to ensure research articles, supposing that editorials, opinions, comments, among others are not expected to receive enough citations to be among the most-cited publications. In this study, we found an opinion article among the most-cited, besides debates, forums, and notes, which were then categorized by the authors according to the classifications considered in this investigation (original, methodological, or review). This is an error that can be attributed to the indexer or editor of the journal. This is a limitation of our study, since other records not indexed as original articles or reviews may have had more citations than the 105 articles included here.
Another limitation is that we used only the Scopus database in data collection. However, this is the database that indexes most journals of public health edited in Brazil, which is why it was chosen. Also, our findings are limited by the fact that we have not assessed the quality of the citations.