| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Special Reports |
1 From the MR Section, La Plana de Vila-Real Hospital, Castellón, Spain (A.M.); Departments of Radiology (L.M.) and Preventive Medicine (D.B.), Dr Peset University Hospital, Valencia, Spain; Institute of History of Science and Documentation Lopez Piñero, University of Valencia-Consejo Superior de Investigaciones Científicas, Valencia, Spain (R.A.); and Cardenal Herrera-Centro Estudios Universitarios, Alfara, Valencia, Spain (P.S.). Received January 31, 2005; revision requested April 7; revision received May 15; accepted June 21; final version accepted July 18. Address correspondence to A.M., C/Pintor Peris Aragó 35-12, Alboraya (Valencia) 46120, Spain (e-mail: amdasit{at}hotmail.com).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
Materials and Methods: After searching the Medline database for articles published between 2000 and 2003, all articles that originated from presentations at the ECR in 2000 and were published in journals indexed in the Journal Citation Report were evaluated. Relative likelihood of full publication depending on the country in which an abstract originated was compared with that of Austria, which is the host country of the ECR. The 
2 test was used to compare publication rates according to country of origin and collaboration. The median and interquartile range of the impact factor of publications according to country of origin and collaboration in the abstract were analyzed by using the Kruskal-Wallis test. The percentage of articles published in journals ranked in the top quartile (top 25% of journals according to impact factor) was calculated.
Results: Of 1020 presentations, 403 articles (39%) with significant (P < .001) variations according to country of origin were identified. Studies originating from the United States had the highest percentage of full publication (62%; relative likelihood, 1.41), median impact factor (4.5), and percentage of articles in the top quartile (54%). Radiology was the top quartile journal, with the most articles published (60%). Impact factor differed according to country of origin (P < .001). Abstracts submitted with collaboration between (a) European countries and countries outside of Europe (other than the United States) and (b) different European countries had the highest publication rates (83% and 52%, respectively; P = .03).
Conclusion: The country of origin of an abstract was useful in predicting the chances of full publication of the abstract, with the United States having the highest publication rate. Authors were more likely to publish an article if it had been prepared with international collaboration.
© RSNA, 2006
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
Collaboration can be defined as the act of working together. Scientific collaboration is a means to advance research and enhance publication capacity (4). Nowadays, it is common for scientists to conduct research in collaboration with their colleagues from different institutions, countries, or disciplines (5). Thus, collaboration between authors and institutions is a variable that can influence the rate of full publication of an abstract from the multitude of oral presentations at major medical meetings.
The European Congress of Radiology (ECR) is an important international meeting, and—in our opinion—it is the most important European annual event in radiology. The purpose of our study, therefore, was to estimate the influence of the authors' country and collaboration on the probability of subsequent full publication of abstracts presented orally at the ECR in 2000.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
Two authors (A.M., R.A.) then jointly estimated the number of Medline-indexed journals publishing articles from orally presented abstracts at ECR 2000 that were also included in the JCR database by scanning a Web site (http://go5.isiknowledge.com/portal.cgi) and registering their impact factor for each year of the period between 2000 and 2003. The observation period for full publication of the orally presented abstracts was restricted to a 4-year interval after ECR 2000. This 4-year interval was selected because at the time of the JCR database search and preparation of this manuscript, the 2004 JCR science edition was not available to enable accurate calculation of the 2004 impact factor of possible journals. In 2000, the ECR was held in March; therefore, articles published in January or February 2000 were excluded from this study. The mean impact factor for each journal between 2000 and 2003 was calculated and expressed as the impact factor value. Journals not indexed by JCR and journals not included in the JCR database for all years from 2000 to 2003 were excluded from the study.
Variables Evaluated
The first variable evaluated was the country from which the abstract was submitted; we considered the submitting country to be the first one listed. Only countries with 10 or more abstracts presented at the ECR in 2000 and more than three published articles from abstracts presented at the ECR in 2000 were included in the analysis as the most representative countries. From these countries, publication rates and percentages of papers published in journals ranked in the top quartile according to impact factor were calculated. The top quartile (top 25% of journals according to impact factor) was selected to minimize differences in the characteristics of articles (basic or clinical research, size of field, publication and citation patterns). Better comparison based on impact factors is possible by using a normalized indicator as the percentage of articles published in journals ranked in the top quartile (7). Furthermore, the publication rates of oral presentations according to the authors' country of origin are largely influenced by the number of inhabitants in specific European countries (eg, in 2002, Germany had roughly 80 million inhabitants, whereas Switzerland had only 7 million inhabitants). Because of interest in the investigation of the scientific productivity of individual countries according to the publication rate per 1 million inhabitants, oral presentations and subsequent publication were adjusted for the number of inhabitants of the countries evaluated.
The second variable evaluated was collaboration between authors of abstracts published as full-text articles. Each article was classified according to five types of collaboration: type 1 (no collaboration, abstract submitted from one institution), type 2 (abstract submitted from different institutions within a European country), type 3 (abstract submitted from different European countries), type 4 (abstract submitted with collaboration between a European country and the United States), and type 5 (abstract submitted with collaboration between European countries and countries outside of Europe other than the United States). For each abstract included in the abstract book of the Proceedings of the 12th ECR (6), collaboration was measured by identifying the countries that participated in the corresponding study. For example, in the abstract entitled "Prostate Cancer Localization with Combined MRI and 3D 1H MR Spectroscopic Imaging: Effect of Hormone Therapy" by Mueller-Lisse et al, which was expanded into a full-text article published in Radiology in 2001 (8), collaboration between physicians in Germany and the United States was identified. Thus, the collaboration in this article was classified as type 4.
The third variable evaluated was the language of the publishing journal. The fourth variable evaluated was concordance between the country in which the published article originated and the country of origin of the JCR-indexed journal that published the subsequent article. It was difficult to assess the country of origin for some journals (eg, the journal Cardiovascular and Interventional Radiology has European, North American, and Asian editorial offices); therefore, we assessed the country of origin of the journal according to the geographic listing in the List of Journals Indexed in Index Medicus (9).
Statistical Analysis
Publication rates according to the country from which the abstract originated were calculated and compared with the Pearson 2 test. When use of the 2 test was inappropriate, the Fisher exact test was used. Relative likelihood of full publication and 95% confidence intervals were estimated for the different countries, with Austria serving as the reference standard, as this is the country in which the ECR is located. Publication rates according to types of collaboration in the abstract were evaluated and compared by using 2 statistics. The median and interquartile range of the impact factor of the journals publishing these articles, according to both the country from which the presentation originated and the type of collaboration, were estimated and compared by using the Kruskal-Wallis test. A variable was considered influential on the basis of likelihood ratio test results at the .05 level. Statistical analysis was performed by using computer software (SPSS, version 11.0; SPSS, Chicago Ill).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
Publication rates according to country of origin (Table 1) show a significant relationship (P < .001, 2 = 62.24) to the subsequent publication of the full-text article. When taking into account the 15 most represented countries, all of which had 10 or more studies presented at the ECR in 2000 and more than three published articles, authors from Germany published the most articles (161 of 343 [47%]; relative likelihood, 1.07; 95% confidence interval: 0.83, 1.38), whereas authors from the United States had the fewest number of abstracts but the highest rate of full publication (13 of 21 [62%]; relative likelihood, 1.41; 95% confidence interval: 0.94, 2.12). Publication rates adjusted for the number of inhabitants of these 15 countries (Table 2) show that only Austria (5.18), Switzerland (2.23), Germany (1.96), Greece (1.41), and Belgium (1.07) had more than one article published per 1 million inhabitants, whereas Korea (0.12), Poland (0.08), Spain (0.07), the United States (0.05), and Japan (0.03) had the lowest adjusted publication rates.
|
|
|
2 = 10.79, P = .03) but does not significantly influence the impact factor (Kruskal-Wallis test, P = .54). Abstracts with type 5 (abstract submitted with collaboration between European countries and countries outside of Europe other than the United States) and type 3 (abstract submitted with collaboration between different European countries) collaboration had the highest publication rates (83% and 52%, respectively).
|
Origin of Abstracts and Journals
There was significant concordance between the origin of an abstract and the location of the journal in which the expanded article was published (2 = 126.96, P < .001). Authors from the United States published 12 (92%) articles in journals published in the United States, German authors published 76 (47%) articles in journals published in Germany, and authors from the United Kingdom published 14 (38%) articles in journals published in the United Kingdom. Authors from the Netherlands, France, Switzerland, and Italy published only one (8%) of 13, one (7%) of 14, one (6%) of 16, and two (5%) of 38 articles, respectively, in journals published in their respective countries.
Table 4 shows the 23 journals ranked in the top quartile according to their impact factor, the country in which the journal is published, and the distribution of the articles. In these journals, 63 (16%) articles originated from 10 different countries. Radiology published the most articles (n = 38, 60%) that originated from a different country. Of the 63 articles, 47 (75%) originated in European countries and were published in U.S. journals, whereas nine (14%) articles that originated in European countries were published in European journals, and six (10%) articles that originated in the United States were published in U.S. journals. Only one (2%) article from the United States was published in a European journal ranked in the top quartile.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
It has been known since the 1950s that fewer than half of the papers presented at medical conferences are published as full-text articles (13,14). The publication rates of papers presented at radiologic meetings range from 9% to 37% (15–18). Our analysis of the publication rate in JCR-indexed journals of abstracts presented at the ECR in 2000 showed the highest published publication rate (39%) from a radiologic meeting.
The country from which the abstract originated influenced the likelihood of its subsequent publication as a full-text article. Arrivé et al (16) compared the publication rates in Medline-indexed journals of original studies that originated in several countries and were presented at the 1995 Radiological Society of North America Annual Meeting. These investigators reported that studies from the Netherlands and Japan had the highest publication rates (59% and 47%, respectively) (16). We compared our findings with the findings of Arrivé et al (16) and found that the rate of full publication of studies from Japan was lower (25%), whereas the rate of full publication of studies from the Netherlands was similar (54%). In our series, studies from the United States had the highest publication rate (62%); however, Germany had the most abstracts (n = 343 [34%]) and published articles (n = 161, [47%]).
Although it must be assumed that authors who presented abstracts had a specific interest in presenting their results not only orally during the ECR but also to the whole scientific community in the form of a full article, caution is required when evaluating the importance of some countries (eg, Korea, Poland, Israel, Spain, Japan, the Netherlands, and the United States) with few presentations at the ECR in 2000 in the subsequent publication of full-text articles. Furthermore, when the publication rate was adjusted for the number of inhabitants of the countries, Korea, Poland, Spain, the United States, and Japan had low publication rates per 1 million inhabitants, whereas Austria—the host country of the ECR—had the highest adjusted publication rate (ie, 5.18).
Journal quality is a characteristic that is difficult to define objectively. Perhaps the most often quoted indicator of quality is the impact factor, which is a measure of the number of times the average article in a journal is cited (19); however, this method does have weaknesses. The choice of citations is subjective, and several variables may influence this selection, such as the type and size of a journal, domain concerned, language of publication, self-citation rate, coding of articles based on their nature, and choice of articles published (20). Nevertheless, the journals with the highest impact factors in each field are attractive targets for many scientific workers. In a citation analysis study of published articles originally submitted to an emergency medicine conference, Callaham et al (3) suggested that the strongest predictor of citations per article was the impact factor of the original publishing journal. Our study shows that countries with the most published articles (Germany, Austria, Italy, and the United Kingdom) had a lower percentage of articles published in the 23 journals ranked in the top quartile according to impact factor when compared with articles from the United States and the Netherlands. These two countries, as well as Spain, had the highest median impact factor and the highest percentage of articles in the top quartile. Few radiologic journals were ranked in the top quartile, and the majority of articles were published in the U.S. journal Radiology.
Multi-institutional and international collaborations are related to research quality. With respect to published articles in which only one institution was indicated in the presentation, we found an increased rate of full publication of articles derived from abstracts in which research was performed with collaboration between two or more institutions located in (a) the same European country, (b) different European countries, and (c) a European country and a country located outside of Europe (excluding the United States). Our study results show that multi-institutional and international collaboration produces an increase in the number of published articles. A possible bias in this result is that it is difficult to assess collaboration of authors from different institutions and countries by looking at the affiliation footnote given in the journals at the time of full publication of an original research study presented at an international medical meeting. Authors from Europe, especially if they are scientifically active, may have worked at a North American institution during their postgraduate scientific education; therefore, they might submit manuscripts with a North American affiliation. Other authors might move from one institution to another during their residency; therefore, they may submit a manuscript that originated from research performed at their former institution with the affiliation of their current institution. Another factor influencing collaboration is the need to share resources. In Europe, hurdles of clinical research include the lack of financial support, which leaves many medical research projects unfunded. Although there are no immediate solutions to this problem, existing resources must be optimized. This aim can be achieved through increased collaboration and improved coordination of existing research structures across Europe (21).
English is the predominant language in contemporary medical research; therefore, the majority of articles were published in English-language journals. It should be noted that articles were published in German-language journals with the second highest frequency; this is probably due to the high number of presentations from Germany and Austria and the several radiologic journals that publish in German, such as Fortschritte auf dem Gebiete der Rontgenstrahlen und der Nuklearmedizin, Der Radiologe, and Ultraschall in der Medizin. However, 73% of articles that originated in Germany were published in English-language journals; the remaining articles were published in German-language journals.
There is a concordance between the country of origin of an abstract and the country of the journal that published the full-text article. In this sense, of the few abstracts that were written in the United States and were presented at the ECR in 2000, nearly all of the derived articles were published in U.S. journals. In this sense, most North American research is presented at meetings in the United States and published in U.S. journals. According to the study of Arrivé et al (16), a total of 1202 presentations at the 1995 Radiological Society of North America Annual Meeting originated in the United States; in contrast, 21 abstracts that originated in the United States were presented at the ECR in 2000. Furthermore, only 37% of abstracts prepared with collaboration between institutions in the United States and institutions in European countries were expanded into full-text articles, whereas 52% of abstracts prepared with collaboration between European countries were published as full-text articles. On the other hand, the majority of articles that originated in nine different European countries and were published in journals ranked in the top quartile were published in journals edited in the United States, chiefly Radiology. This finding suggests that European investigators who tend to publish their articles in journals with the highest impact factor tend to select journals that are edited in the United States.
Our study had some limitations that warrant further consideration. One factor influencing the rate of subsequent publication of an abstract is publication bias. Studies with statistically significant results are published more frequently than studies without significant results in prestigious journals (ie, journals with a high impact factor) (22). We did not specifically evaluate this variable in our survey, so we cannot comment on its importance. Moreover, we determined publication status on the basis of our PubMed search; therefore, articles published in journals not included in the Medline database were considered unpublished. In this sense, articles not published in English are surely underrepresented, as PubMed focuses on English-language journals (23). Many European radiologic journals, especially those that do not publish in English, are not listed in the Medline database. The absence of these journals can lead to a selection bias. Thus, improved strategies are needed to identify the results of unpublished, as well as published studies (22). Another limitation of our study was the consideration that the first author of a presentation represents the country from which a presentation originates. This can be largely influenced by a variety of factors (eg, a move from one institution to another during residency, sabbatical leave for a research project).
Although the most informative and highest quality studies are published in peer-reviewed journals, some comments on unpublished abstracts must be made. Reasons for not publishing abstracts are multifactorial, but the findings of several studies (24,25) showed that the main reason studies were not published was lack of time. It seems reasonable not to expect that all research presented at meetings be submitted for publication for a variety of reasons, including the fact that there is not enough space in high-quality journals, peer review is much less rigorous at the meeting level than at the journal level, and much of the material presented at meetings is simply not of high enough quality or of new or great enough importance to be published. For these reasons, the rate at which abstracts are published as full-text articles can be considered a measure of the quality of scientific meetings.
In conclusion, approximately 40% of abstracts presented orally at the ECR in 2000 were subsequently published in JCR-indexed journals, with the U.S. journal Radiology having the highest number of papers published in journals ranked in the top quartile according to impact factor. The author's country and international collaboration (excluding collaboration between European countries and the United States) influenced full publication. Studies originating in the United States had the highest percentage of full publication, median impact factor, and percentage of articles in the top quartile.
|
ADVANCES IN KNOWLEDGE |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
|
FOOTNOTES |
|---|
Abbreviations: ECR = European Congress of Radiology • JCR = Journal Citation Report
Author contributions: Guarantor of integrity of entire study, A.M.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; manuscript final version approval, all authors; literature research, A.M., R.A., P.S.; statistical analysis, A.M., L.M., D.B.; and manuscript editing, A.M., L.M.
Authors stated no financial relationship to disclose.
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
Related Article
This article has been cited by other articles:
|
|
 |
|
  L. Arrive, M. Lewin, P. Dono, L. Monnier-Cholley, C. Hoeffel, and J.-M. Tubiana Redundant Publication in the Journal Radiology Radiology, June 1, 2008; 247(3): 836 - 840. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Arrive and A. Miguel-Dasit Declining Number of Publications by American Authors in Radiology Radiology, January 1, 2008; 246(1): 330 - 331. [Full Text] [PDF]
|
|
|
|
|
|
M. A. Pawlak, J. Krejza, A. Miguel-Dasit, and L. Marti-Bonmati Scientific Presentations at Radiologic Meetings, Subsequent Full-Text Publications, and Authors' Country of Origin Radiology, January 1, 2008; 246(1): 331 - 332. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
