Objectives To conduct a comprehensive mapping analysis to the scientific literature published in football aiming to identify the areas of bigger interest and potential for further exploration.
Methods The data were obtained by a search conducted on the Web of Science. Articles were listed based on citation frequency. We used an open-source bibliometrix R-package for the comprehensive bibliometric analyses.
Results The number of citations per article ranged from 251 to 869 (median 323; IQR 125). The yearly number of citations ranged from 8 to 54 (median 26; IQR 11). Most of the articles (76%) were of level III of evidence, 10% were level II and 14% were level IV. Within the top 50 most-cited articles, 40 articles were original research (37 observational and 3 experimental studies), 9 were review articles and 1 was a thesis. From the 40 original research articles, 50% involved elite players, 73% were exclusive to male players and 80% involved adult players only. The topic area with the highest number of articles was sports medicine (44%), followed by training and testing (32%), performance analysis (14%) and physiology (10%). No study within the top 50 was devoted to biomechanics, nutrition, sport psychology, coaching or social sciences.
Conclusions The lack of experimental studies within the top 50 most-cited articles in football clearly underpins how far we still are from establishing the theoretical and methodological guidelines for the applied science and medicine in football.
- bibliometric analyses
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/
Statistics from Altmetric.com
Given the increasing number of academic publications in football, even the most fervent reader will appreciate it is impossible to keep up to date with all publications. Moreover, the football community is flooded with empirical contributions from different sources and networks that tend to mislead and fragment knowledge and evidence. One simple tool to evaluate the quality of research conducted is the number of citations. Bibliometric analyses are used by organisations, institutes and universities to evaluate the quality of the research output. This type of analysis is widely used in other disciplines like medicine and its sub-specialties.1–4
In sports, and especially in football science, bibliometric analysis might first assist in quantifying the quality of studies and also identify areas that have attracted most of the interest until now. As an extension to this bibliometric analysis, we will be able to identify the topics that have received less scientific attention so far. The lack of attention, indicated by the low number of citations, might indicate relatively lower studies’ quality and/or less interest by the community.
In the current manuscript, we used automated software workflows to perform comprehensive mapping analysis to the scientific literature published in football. The bibliometric analyses were used to establish the top 50 most-cited articles in medicine and science in football, aiming to explore the impact of the most prominent publications, journals, authors and networks.
The data in this study were obtained by a search conducted on the Web of Science (Clarivate Analytics, USA). There were no restrictions applied based on availability of abstract, study type, language and human versus non-human research objects or any kind of time limitation. Keywords ‘soccer’ and ‘football’ were searched in terms of topic (including four sections: paper title, abstract, author keywords and KeyWords Plus) based on the full collection (10 April 2018). Our search produced 41 191 published articles between 1888 and 2018. Thereafter, they were listed based on citation frequency from the highest to the lowest. All articles focused on football or with a population of football players were considered eligible for analysis, but if no independent data were presented for football players, the article was excluded.
The 50 most-cited articles were reviewed by two investigators (JB and PF) and the following information was extracted according to their specific characteristics: (1) year of publication, (2) journal title, (3) number of citations, (4) number of annual citations, (5) authorship and (6) level of evidence (I–V) from Oxford Centre for Evidence-Based Medicine 2011.5
The articles were further independently characterised by two investigators (JB and PF) and discussed in the case of discrepancies. Then, a third investigator (ATS) re-analysed all the articles, and a discussion was employed until a consensus was achieved. Each article was characterised according to category (original research, review article, case study, short communication/technical report, letter to the editors, point–counterpoint, editorial and thesis), main research topic area (physiology, biomechanics, nutrition, training and testing, sports medicine, performance analysis, sport psychology, coaching and social sciences) and type of study (observational, experimental and review). Also, sex (male, female and both), competitive level (elite, non-elite and both) and age group (youth, adult, all) were extracted for the original research articles (ie, observational and experimental studies). For the purpose of our study, competitive levels were defined as elite (when participants were classified as professional, elite or first division) and non-elite (for amateur, non-elite, college or high-school players), and age groups as youth when <18 years old and adult when >18 years old. When the information was not evidently provided in the article, these categories were considered unclear.
The data analysis was performed using R.6 An open-source bibliometrix R-package for performing comprehensive bibliometric analyses was used.7 Also, a network creation for bibliographic collaboration was performed. A scientific collaboration network is a network where nodes are authors and links are co-authorships; it is one of the most well-documented forms of scientific collaboration.8 An author collaboration network can be obtained computing Bcoll=A×A′, where A is a Document×Author matrix. Element bij indicates how many collaborations exist between authors i and j. The diagonal element bii is the number of documents authored or coauthored by researcher i (for details, see Aria and Cuccurullo7).
The distribution of a parameter (eg, number of citations) was characterised by the median (IQR). The χ2 test was used to compare two categorical variables. To test the hypothesis if three or more samples characterised by their medians originated from the same distribution, the Kruskal-Wallis test was conducted. The significance level was set to p value <0.05.
The top 50 most frequently cited articles in medicine and science in football are shown in table 1. All articles were published in English. The number of citations ranged from 251 to 869, with a median of 323 and an IQR of 125. The number of citations per year ranged from 8 to 54, with a median of 26 and an IQR of 11. Most of the articles (38) were of level III of evidence, while five articles were level II, and seven articles were level IV (χ2=41.08; p<0.001). No significant differences (χ2=1.57; p=0.46) were found between the citation medians of the articles and the level of evidence.
Of the top 50 most-cited articles, 40 articles were original research (37 observational and three experimental studies), nine were review articles (eight narrative reviews and one meta-analysis) and one article was a thesis encompassing 14 original articles and a summarising review (χ2=50.92; p<0.001). No significant differences were detected for the number of citations according to the type of article (χ2=0.751; p=0.687; figure 1). The distribution of original research articles varied according to sex (χ2=49.4; p>0.001), age group (χ2=66.2; p<0.001) and competitive level (χ2=15.8; p=0.001) of the participants.
The majority (n=29) of the 40 original research articles investigated male players, and only four studies were exclusively dedicated to female players. Six studies included both male and female players, and the sex of the participants was unclear in one study. However, no significant differences were detected for the number of citations according to the sex of the studied population (χ2=1.527; p=0.676).
Similarly, 32 original research articles investigated adult players, and only six articles were dedicated to youth players. One study included both adult and youth players, and the age group was unclear in one study. No significant differences were detected for the number of citations according to the age group of the studied population (χ2=3.066; p=0.382).
Half of the original research articles (n=20) investigated elite players, and 10 articles were conducted with non-elite players. Seven articles included both elite and non-elite players, and the competitive level was unclear in three studies. No significant differences were detected for the number of citations according to the competitive level of the studied population (χ2=4.599; p=0.204).
The topic area with the highest number of articles (χ2=15.12, p<0.01) in the top 50 most frequently cited articles was sports medicine (n=22), followed by training and testing (n=16), performance analysis (n=7) and physiology (n=5). No differences were detected for the number of citations per topic area (χ2=1.133; p=0.769; figure 2).
Within the original research articles, the samples sizes ranged from 14 to 2376 players. Eleven articles involved less than 30 players, and 18 articles had a sample size above 50 participants. The sample size was unclear in five original research articles. The majority (n=33) of the original research articles were exclusively dedicated to football, but seven articles were multisports investigations.
The top 50 most frequently cited articles were published in 17 journals (table 1), with 68% of the articles published by five journals: Journal of Sports Sciences (n=8), American Journal of Sports Medicine (n=8), Medicine and Science in Sports and Exercise (n=7), British Journal of Sports Medicine (n=6) and International Journal of Sports Medicine (n=5). The articles were published during the time period 1983–2011 (figure 3). The oldest article was published in 1983 in Medicine and Science in Sports and Exercise, and the most recent ones in 2011 in British Journal of Sports Medicine and American Journal of Sports Medicine.
Overall, the top 50 most-cited articles encompassed 139 authors in total, with an average number of 2.8 authors per article. There were two single-author articles. The maximum number of coauthors in an article was 9. The list of authors with three or more articles in the top 50 is presented in table 2. Additionally, a clustered-network of coauthorship collaborations within the 18 authors with three or more articles in the top 50 is presented in figure 4. Seven clusters of co-authorship collaborations were identified.
Over the last decades, the scientific literature available in football has dramatically increased both in terms of quantity and quality. However, to our knowledge, this is the first bibliometric analysis conducted in football. Here, we show that the number of citations for the 50 most frequently cited articles in medicine and science in football ranged from 251 to 869.
A bibliometric analysis is an extraction of statistics on journal articles in a research field. Its structure resembles the design of a systematic review.9 Bibliometric analyses are commonly used to measure the impact of scientific articles, but should not be considered as the standard reference of scientific quality of the investigations. Citation rankings may also provide an estimate for trends for the research undertaken worldwide in a specific research subject (eg, football). We observed that 80% of the articles presented in the current ranking were original research, but only three articles had a clear experimental design. Observational studies (eg, notational analysis of performance) have a long tradition in football research, aiming the identification of critical events (often referred as performance indicators) that are thought to be the keys to success in the sport.10 In fact, observation studies are essential for a comprehensive awareness of the distinct individual and team performance components. Notwithstanding, the lack of experimental studies within the top 50 most cited articles in football clearly underpins how far we still are from establishing the theoretical and methodological guidelines for the applied science and medicine in football.
Nine review articles, but only one meta-analysis, were included in the current ranking. Meta-analyses characterise and combine findings of prior studies in order to increase statistical power, provide quantitative summary estimates, and identify data gaps and biases.11
Some of the limitations for conducting scientific studies in football are well known: it is difficult to have access to large sample sizes of top-level players, longitudinal training studies are arduous and laborious, and studies that address and replicate appropriate real-world methodological issues are rare. However, despite the important role of narrative review articles, the debate within the community dealing with medicine and science in football might still be far from high levels of evidence.
Still, the theoretical knowledge, methodological approaches and professional practice in football are supported by comprehensive and complex contributions from a variety of sources.12 In this line, the input of individuals involved in sports medicine, both from a clinical and scientific perspective, should be highlighted. It is noteworthy that 44% of the articles presented in the current ranking were related to sports medicine. The other topics main topic areas were training and testing, performance analysis and physiology, but no study within the top 50 most-cited was devoted to biomechanics, nutrition, sport psychology, coaching or social sciences.
The articles presented here were all published between 1983 and 2011. Online databases have limitations in tracking older articles, especially for articles published in alternate media, textbooks or meeting symposia,4 which may have been missed in the current analysis. In addition, it can also be argued that old data are of limited value in the current rapidly changing sports environment. It should be noted that the impact of recent studies could not be captured with the current analysis. Any article needs some time after publication to receive enough number of citations. However, this is common in all research fields and does not negate the trend and conclusions presented here. We might expect a relative emergence of systematic research analysis and experimental studies in medicine and science in football over the last few years, underpinning the recent growing interest in those areas and, at the same time, highlighting the need for more research in the future. Still, the current study may also be limited by other research publication dynamics, which can include publication biases, as language, predominance of positive results, journal preferences when citing and self-citation.13 14
In the current manuscript, we present a network of authors’ collaborations within the scope of medicine and science in football. This network underpins the most prolific scientific collaborations within medicine and science in football. Five main clusters of network collaborations were identified. We hope the current manuscript can establish an historical, comprehensive analysis to the most influential scientific literature published within the area of medicine and science in football, helping researchers, sports scientists, clinicians and practitioners on developing new insights to the original theoretical and methodological understanding of the game of football.
Contributors JB and PF designed the study, collected and analysed the data, and wrote the manuscript. GPN and ATS analysed data and wrote the manuscript.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient consent Not required.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.