Article Text

Patterns of co-authorship in international epidemiology
1. K H Jacobsen
1. Dr K H Jacobsen, Department of Global & Community Health, College of Health and Human Services, George Mason University, 4400 University Drive 5B7, Fairfax, VA 22030, USA; kjacobse{at}gmu.edu

## Abstract

Background: The objective of this study was to identify current patterns of international collaboration based on authorship of epidemiological articles.

Methods: All articles published in nine high-impact public health journals in 2006 were read, and information about the study country and the authors’ countries of affiliation was extracted. All countries were assigned to an income level using World Bank classifications. This paper presents results for the 1686 articles that focused on a single study country.

Results: International collaboration is common, but the dominant partnerships vary by the income level of the study country. 74.2% of articles reporting on research conducted in low- and middle-income study countries involved co-authors from two or more countries, and nearly all of these international collaborations included co-authors from both low/middle-income and high-income countries. Only 13.0% of studies based in high-income countries involved co-authors from two or more countries, and the majority of these studies involved co-authors with affiliations solely from high-income countries. More than 90% of articles from both low/middle- and high-income study countries included at least one co-author from the study country.

Conclusion: There is a high rate of local co-authorship in both low/middle- and high-income study countries. Most articles that focus on high-income countries include only authors from high-income countries. Most articles that focus on low- and middle-income study countries are “north-to-south” international collaborations that include co-authors from both low/middle- and high-income countries. “South–south” partnerships are rare.

## Statistics from Altmetric.com

International collaboration is defined as research conducted by scholars from two or more countries working together on a project. These international collaborative efforts may be relatively simple—such as two individual scientists working together on a project—or may take on more complex forms, such as institutional partnerships, research networks, public–private partnerships and multilateral international research centres.

International collaborative research offers a number of potential advantages, including greater access to resources such as expertise, equipment and funding; broader visibility for the project; and the potential for increased efficiency, speed and productivity.14 International co-authorship has also been found to increase the number of times an article is cited.5 6 Given these possible benefits, it is not surprising that international research collaboration is becoming more common. The proportion of articles from the biomedical sciences that include co-authors from two or more countries has been increasing in recent decades, as shown in evaluations of the literature of the 1970s, 1980s and 1990s.58

References to the global “north” and global “south”, regions defined by both income level and geography, are often used to describe international partnerships. This language is imperfect, in part because some countries in the “geographic south” (including Australia and New Zealand) are considered to be part of the “global north”, but the terms are still useful for describing the nature of current research networks. International collaborations include “north–north” exchanges between academics from higher income countries, “south–south” collaborations between researchers from lower income countries, and “north–south” partnerships of scientists from countries in different income strata. Four categories for collaboration can be created by dividing “north–south” partnerships into two directional groups: “north-to-south” research that focuses on a country in the “south” and “south-to-north” research that focuses on a country in the “north”.

The goal of this study was to identify current patterns of authorship and international collaboration in epidemiology. Authorship characteristics for all of the nearly 1700 articles published in nine high-impact public health journals in 2006 that focused on one study country were examined. The study country was defined as the one country on which the article focused, and author countries were identified based on affiliations listed in the journal articles. Results of the evaluations of authorship characteristics are reported by country income level and by world region. Including the study country in the analysis rather than just authorship allows for identification of “north–north”, “south–south”, “north-to-south” and “south-to-north” collaboration.

## METHODS

### Journal selection

Journal Citations Reports from the ISI Web of Knowledge9 were used to identify highly rated journals with a focus on epidemiology and/or public health. The nine journals included in this analysis are all listed as “Public, Environmental & Occupational Health” journals, published more than 125 articles in 2005, had an impact factor of 1.500 or greater and had been cited at least 2000 times in total by the end of 2005. The journals that met these criteria are the American Journal of Epidemiology, American Journal of Public Health, American Journal of Tropical Medicine & Hygiene, Epidemiology & Infection, International Journal of Epidemiology, Journal of Clinical Epidemiology, Journal of Epidemiology and Community Health, Transactions of the Royal Society of Tropical Medicine and Hygiene and Tropical Medicine & International Health.

### Limitations

The criteria for selecting journals for inclusion in this analysis were designed to yield a sample of journals that publish a large number of articles each year (a minimum of about 10 articles or more per month), cover topics of interest to a wide range of international readers and are frequently cited. The highly ranked English-language public health journals identified by these criteria are not intended to be representative of the entire public health literature. The national and local publications that make up the majority of public health journals are underrepresented. The goal of this analysis, however, was not to determine what proportion of all public health and epidemiology articles have international co-authorship, but to determine what proportion of articles published in highly ranked international journals involved international co-authorship. Regional periodicals are usually tailored to local interests rather than international themes, and may involve less international collaboration. Also, the journals that met the inclusion criteria included several tropical medicine journals, and that led to a much higher proportion of studies focused on low- or middle-income countries (about 40%) than was found for previous studies of articles in social and biomedical journals.8 1015 To adjust for this potential oversampling of studies from low- and middle-income countries, authorship characteristics will be reported by the study country’s income stratum or world region, and no summary statistics for the entire sample will be reported.

### Article eligibility

All articles published in these nine journals in 2006 were examined. After eliminating editorials, letters to the editor, invited responses to articles, book reviews and other special features, there were 1991 remaining articles, of which 1686 (84.7%) were based on research conducted in only one country, 89 (4.5%) were based on research in more than one country and 216 (10.8%) were theory and review papers that were not country-specific. This analysis focuses on the 1686 studies that were conducted in only one country. More than 120 study countries are included in the data set.

### Author affiliation

All of the included journals provide a list of the institutional affiliation or affiliations of each author. For each article, the country or countries of affiliation of the first author were noted along with a separate list of countries of affiliation for all authors (including the first author). A manuscript was said to be the result of an international collaboration if the co-authors of the paper had affiliations in two or more countries. Reliance on institutional affiliation means that some information about authors’ affiliation is incomplete or inaccurate. In some cases, researchers from one country who are temporarily working or studying in another country may be listed as having an affiliation only in the host country, and this might lead to an underestimation of the proportion of studies that involve international collaboration. In other cases, listing of multiple affiliations for individual authors may lead to the appearance of greater collaboration than truly exists, if authors with multiple affiliations claim a secondary affiliation in the study country when their primary affiliation is in another country. This is a limitation of the current standard for reporting of affiliation and adjustments for it cannot be made in this analysis.

## RESULTS

### Country income level

The World Bank uses estimates of gross national income (GNI) per capita in US dollars to assign countries and territories to income groups. For 2006, countries with a per capita GNI of less than or equal to $11 115 were considered to be low- and middle-income countries (LMICs) and countries with a per capita income greater than$11 115 were considered to be high-income countries (HICs). LMICs can be further divided into lower income countries with a GNI less than or equal to $905, lower middle income countries with a GNI between$906 and $3595 and upper middle income countries with a GNI between$3596 and \$11 115.16 These World Bank classifications were used to assign each study country and author country to an income group. Studies from three territories not included in the World Bank data—French Guiana, Marshall Islands and Tuvalu—were categorised as LMICs. The 95 (5.6%) first authors who listed affiliations in more than one country were considered to be from both an LMIC and an HIC if the countries of affiliation were in different income groups.

International collaboration was much more common in LMIC study countries (74.2%) than in HIC study countries (13.0%) (table 1). Increased collaboration is associated with an increased number of co-authors: the mean number of authors was significantly higher for LMICs (seven) than for HICs (five). There was a high rate of participation by “local” authors in both LMICs and HICs, with more than 90% of studies from all income levels including at least one co-author from the study country, but studies based in LMICs were much more likely to have a co-author from another country. The majority of studies focusing on an LMIC included at least one co-author from an HIC, but less than 1% of studies focusing on an HIC included any co-author from an LMIC. In other words, researchers from both LMICs and HICs participated as co-authors on papers about LMICs, but in most cases only researchers from HICs served as co-authors on papers about HICs. Studies conducted in LMICs were also less likely to have a first author from the study country.

Table 1 International collaboration by country income level

There are five possible combinations of authors by income level (table 2). The majority of papers from HICs included co-authors from only one high-income country (87.0%) and the majority (71.8%) of papers from LMICs involved “north-to-south” international collaborations. The most common form of international collaboration in HIC study countries was “north–north” collaboration, which accounted for 12.2% of all studies focused on HICs and 93.9% of collaborations conducted in HIC study countries. In other words, international collaborations tend to be fairly unidirectional: “north–north” and “north-to-south” collaborations are common, but “south–south” and “south-to-north” partnerships are rare.

Table 2 Contributors by study country characteristic

### World region

The World Health Organization defines world regions that group countries by geographic, cultural and income characteristics: sub-Saharan Africa (AFRO), the Americas (AMRO), North Africa/Middle East (EMRO, the Eastern Mediterranean region), Europe (EURO), South-East Asia (SEARO) and East Asia/Oceania (WPRO, the Western Pacific region).

Studies based in higher income world regions had lower rates of international co-authorship than lower income world regions (table 3). There was a high rate of participation by “local” authors in all regions. The vast majority of papers in each region included at least one co-author from the study country. Researchers from higher income regions were more likely than researchers from lower income regions to contribute to papers outside their home regions. For example, the rows of table 3 show that less than 1% of papers that focused on a European study country included a co-author from SEARO, EMRO or AFRO, whereas the columns show that European co-authors contributed to more than 30% of papers in SEARO, EMRO and AFRO. In contrast, more than 60% of papers focusing on an African study country included co-authors from EURO, while less than 1% of papers in EURO, AMRO and WPRO included a co-author from Africa. Studies conducted in lower income regions were less likely than studies from higher income regions to have a first author from the study country. As was seen in the income-level analysis, the most common forms of international co-authorship involve “north-to-south” and “north–north” collaborations. “South–south” and “south-to-north” collaborations are rare.

Table 3 International collaboration by world region

## DISCUSSION

This examination of epidemiological articles published in 2006 provides a more precise and nuanced classification of international research partnerships. Several key findings emerge.

First, there is a very high level of “local” co-authorship within all income groups and world regions. More than 90% of studies included at least one co-author from the study country. Previous studies that found low rates of authorship by researchers from low-income countries did not control for the study country,1719 and were therefore unable to observe the high rates of participation by researchers from low-income countries on papers reporting on low-income countries. However, the disaggregated data do show that researchers from low-income countries are less likely than researchers from high-income countries to serve as first author, as has been shown by previous studies.17 18 20

Second, the dominant authorship patterns differ by the income level of the study country. The majority of articles reporting on research conducted in low- and middle-income study countries are “north-to-south” partnerships that involve co-authors from countries with different income strata in different world regions. The vast majority of articles reporting on research conducted in high-income study countries have authors only from the study country.

Third, international collaborations are fairly common and usually take one of two forms at present: “north–north” or “north-to-south.” Researchers from higher income areas participate in the majority of projects conducted in lower income study countries, but researchers from lower income areas rarely participate in projects outside their home countries.

#### What is already known on this subject

International collaboration on research is becoming more common, and the value of collaboration is broadly recognised.

• There is a high rate of “local” co-authorship in both low- and high-income study countries.

• Currently, the most common types of international collaborations are (1) “north–north” partnerships involving co-authors solely from high-income countries in the “global north” and (2) “north-to-south” partnerships with co-authors from different income strata conducting research in a low- or middle-income study country.

• “South–south” and “south-to-north” partnerships are rare.

To better conceptualise contemporary practice, fig 1 describes four stages for the globalisation of international collaborative research. In stage 1, most international collaborations are “north–north” partnerships. In stage 2, the two most common types are “north–north” and “north-to-south” collaborations. A likely next step, shown as stage 3, is that reduced barriers to information exchange21 and the “capacity building” efforts of granting agencies, training programmes and governmental initiatives will lead to an increase in the number of independent scholars from lower income countries and the number of “south–south” collaborations that explore issues of importance primarily to low- and middle-income countries.22 23 Stage 4 represents the globalisation of international collaboration that might occur as the number of researchers from low- and middle-income countries who have management experience and technical expertise grows and as north–south partnerships mature.24 25 In 2006, the two most common types of international co-authorship were “north–north” and “north-to-south” collaborations, and there were almost no publications produced by “south–south” or “south-to-north” collaborations. This indicates that the state of international collaboration in 2006 is best described as stage 2 in the model.

Figure 1

Model for globalisation of international collaborative research.

This study provides baseline information about the nature of international collaborative research in 2006. Including characteristics of both authors and study countries allows for a more accurate depiction of authorship demographics. Two key points have been clarified. First, researchers from low-income countries are serving as co-authors on nearly every paper that focuses on a low-income country. Second, the most common types of international collaboration at present are “north–north” and “north-to-south” partnerships. Globalisation might lead to changes in the dynamics of international collaboration, and future studies of authorship demographics will be able to compare their findings with the data presented here.

## Footnotes

• Funding: None.

• Competing interests: None.

## Request Permissions

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.