Use of data sets collected by surgical collaborating groups Pernille Skou Gaardsted 1 , Henrik Sehested Laursen 2 , Simon Hosbond Poulsen 3 , Hubert Krzyslak 3 and Alaa El-Hussuna 4 ,* on behalf of the OpenSourceResearch Collaboration 1 Medical Library, Aalborg University Hospital, Aalborg, Denmark 2 Medical Library, Regional Hospital Central Jutland, Viborg, Denmark 3 Aalborg University, Aalborg, Denmark 4 OpenSourceResearch Organization, Aalborg, Denmark *Correspondence to: OpenSourceResearch Organization, Aalborg, Denmark (e-mail: alaanewemail@gmail.com) Dear Editor BJS has encouraged collaborative authorship debate 1 . In the past three decades, there has been an increase in publications with large author collaborations where authors come together in so- called collaboratives. The creation of data set(s) by a study group can be very time-consuming. It comprises collection, cleaning, curating, and storage of data, as well as writing up and publishing research based on these data sets. Data sets generated by groups/ collaboratives might, however, be underused 2,3 . The aim of this study was to investigate how surgical collabo- ratives use their data sets. PubMed was searched for articles prepared by study groups in the form of corporate authorship between 2001 and 2018. This was done using the corporate author (CN tag field) in PubMed. Articles in which surgical procedures were designated as a topic were identified using a list of Medical Subject Heading (MeSH) terms. The results were then screened by all the authors inde- pendently using predefined inclusion and exclusion criteria. Details of the search method are explained in Appendix S1. The search in PubMed resulted in 10 579 hits after removing duplicates and applying inclusion criteria. A manual search was done to exclude all corporate authorship studies that investi- gated medical subjects or catheterization procedures, and single- cohort studies, that is those with no well established database reported in the methods section. This resulted in 249 corporate authorship groups covering a wide range of surgical specialties, such as colorectal, orthopaedics, hepatobiliary, paediatrics, gy- naecology, and bariatric surgery. Some 141 of the collaboratives/collaborating groups (56.6 per cent) had only one publication using the data set they collected, whereas 175 (70 per cent) had only one or two publications. Seven collaboratives had more than 10 publications based on their data sets (Table 1). There was a trend over time towards an increase in the number of publications by collaboratives as well as the number of authors per article. Most of the collaboratives (45 per cent) published their studies using both open and conven- tional access; however, 17.3 per cent chose to publish all their studies as open-access articles. Publications from two corporate authorship groups (Danish Colorectal Cancer Group and Dutch Colorectal Cancer Group) were further investigated. The choice of these two groups was based on previous knowledge that both have well established databases from which many publications have emerged. A manual search in PubMed, Google Scholar, Embase, and web- sites of these two groups was performed. This analysis showed that PubMed web-based search algorithms missed some of the publications by the two groups. Some reasons for this might be: No mention of collaboration/collaborative group in the ti- tle; no mention of surgery/surgical intervention under anaes- thesia; no mention of a multicentre prospectively collected data set; or PubMed-related reasons (such as absent field tags, classifications). Many multicentre studies used the term ‘register’ or ‘database’ to describe their data set, although the data were collected only in relation to one study. It was thus difficult to distinguish be- tween a single prospective cohort study and a single publication based on a data set/register. Despite some limitations, the present results may improve reporting of collaborative data sets in the title of studies, and stimulate the creation of new PubMed indexing methods to regis- ter collaborative authorship. The study will hopefully draw atten- tion to the potential problem of wasted resources, encourage collaboratives to share their anonymized data, and guide the con- struction of new data sets with data harmonization across stud- ies to allow pooling. In this respect, OpenSourceResearch might be an option 4 . More than half of the data sets collected by surgical collabo- ratives were used only once. This might be a waste of resources taking into consideration the significant amount of resources used during the data collection process. Better tagging of collab- orative authorship will encourage studying the efficiency of Received: October 07, 2021. Accepted: October 24, 2021 VC The Author(s) 2021. Published by Oxford University Press on behalf of BJS Society Ltd. All rights reserved. For permissions, please email: journals.permissions@oup.com 2 BJS, 2022, 109, e3–e4 https://doi.org/10.1093/bjs/znab403 Advance Access Publication Date: 28 November 2021 Research Letter Downloaded from https://academic.oup.com/bjs/article/109/1/e3/6445205 by guest on 20 December 2023