From Presentation to Paper: Assessment of Successful Abstract Publications in Emergency Medicine Over a Five‐year Period

Le résumé fourni par la source

Abstract presentations are a mechanism to present and discuss up-to-date research with attendees. However, dissemination of research findings is limited when relying on conference presentations in isolation. While some conference proceedings may be published or available online, they are not as easily found as peer-reviewed publications and many are not indexed in major search databases.1 Additionally, abstracts often provide limited details about study methodology, which can make it challenging to adequately assess the study limitations. This can impact the understanding of current research in our field and lead to publication bias in systematic reviews if abstracts are missed or excluded. Prior studies have reported publication rates ranging from 21 to 45%.2-8 However, most of these studies were limited to only a single conference over a limited number of years, with the majority performed in the 1990s and early 2000s. Since then, emergency medicine (EM) has significantly grown as a field and is increasingly developing as a specialty within other countries around the world. Therefore, there is a need to better understand the worldwide publication rates of EM conference abstracts. This cross-sectional study sought to determine the percentage of abstracts that were subsequently published in a peer-reviewed journal. Prior to the study, we reviewed the websites of national and international EM societies to identify which held an annual scientific assembly during which abstracts were presented. We also searched Google and all EM journals for additional conferences with abstract presentations. To be included, the conference must have published the abstract or had the list of abstracts available online and must have hosted at least one conference between 2011 and 2015. We intentionally selected 2015 as the upper limit to allow sufficient time for articles to be published, because previous studies have demonstrated that most articles are published within 4 years of abstract presentation.4, 5 This study was deemed exempt by the institutional review board at Rush University Medical Center. Two investigators independently searched PubMed and GoogleScholar for each abstract using a combination of keywords from the study and author names to identify any publications resulting from the abstract. A publication was linked to the abstract if the specific conference was mentioned in the footnotes of the manuscript or if the studies shared substantial similarities (e.g., study designs, time periods, results, authors). We developed a data extraction tool, which was piloted and modified in accordance with pilot data. To enhance reliability, the investigators were trained on the protocol and check-in meetings were performed to assess questions and issues. To reduce bias, two investigators independently dual extracted the data with discrepancies resolved by a third investigator. The following data were extracted into the data collection form for all included studies: abstract first author, abstract title, conference name and year, manuscript first author, journal name, publication year, first author country, total number of manuscript authors, and study design. When fewer than five manuscripts were published from a given country or study design, they were placed into the “other” category. Data extraction was completed on January 11, 2020. Data were presented primarily as descriptive statistics including percentages. Continuous data were visually evaluated for normality using stem-and-leaf plots and statistically using the Shapiro-Wilk test. In cases where data were not normally distributed, median with interquartile range (IQR) was reported. Statistical analyses were performed using Microsoft Excel (Redmond, WA) or Statistical Package for the Social Sciences (SPSS, Inc., Armonk, NY), version 22.0. We identified 26 conferences comprising 11,043 abstracts, of which 3,974 (36.0%) were published as manuscripts in peer-reviewed journals (Table 1). The number of abstracts published per year did not significantly differ over time. Published manuscripts had a median of 6 authors (IQR = 4). The first author was the same in both the abstract and the manuscript in 3,001 cases (75.5%). The majority of manuscript first authors were located in the United States (67.0%), followed by Canada (11.5%), Australia (4.8%), Turkey (3.6%), South Korea (1.7%), and Denmark (1.7%; Data Supplement S1, Table S1, available as supporting information in the online version of this paper, which is available at http://onlinelibrary.wiley.com/doi/10.1111/acem.14152/full). Fifty-eight percent (2,302/3,974) of manuscripts were published in an EM journal. The most common study designs were prospective observational (36.4%), retrospective (35.3%), cross-sectional/survey (11.4%), and randomized controlled trials (8.4%; Data Supplement S1, Table S2). The median time from abstract presentation to manuscript publication was 2.0 years (IQR = 1.0 years). Thirty-seven of 3,974 abstracts (0.9%) were published prior to the conference. While 95.7% (3,804/3,974) of abstracts were published as papers within the first 4 years, 3.3% (133/3,974) were published beyond that time period with two published 8 years after the abstract (Data Supplement S1, Table S3). When compared with prior studies, our study provides updated data and is one of the most comprehensive to date, comprising 26 different conferences over a 5-year period. As a result, we were able to obtain data on 11,043 abstracts, which is substantially larger than all prior studies. Previous studies have demonstrated variable publication rates over time. Weber et al.2 evaluated 492 abstracts from the 1991 SAEM scientific assembly and reported a 45% publication rate, while Li et al.3 evaluated 2,054 abstracts from the 1997 and 1999–2001 SAEM assemblies and reported a 35% publication rate. Both publication rates were lower than our publication rate for the same conference from 2011 to 2015 (48.5%). Walby et al.4 evaluated 207 abstracts from the Australasian College for Emergency Medicine between 1995 and 1998 and found a 35% publication rate, while we identified a 45% publication rate for this conference. This suggests an increase in publication rates when compared with one to two decades prior. Other studies have reported publication rates ranging from 21% to 33% among various conferences.5-8 Despite inclusion of international conferences, the majority of publications were from countries where English is the first language, suggesting that nonnative English-speaking countries may be underrepresented in the scientific literature. This aligns with prior literature showing that the majority of publications were from countries where English is the first language (e.g., Canada, England, United States, Australia)6, 7 as well as the work from Ehara and Takahashi,9 who found that acceptance rates were significantly lower among countries in which English was not the primary language (29.1% vs. 40.3%). One reason this may occur is greater difficulty in conveying the message due to language barriers. However, EM is also a newer field in many of these countries and the difference may also reflect the reduced number of mentors and dedicated support as the field has not yet fully matured in many of these areas. The lower publication rate has implications for the external validity and applicability of studies to other countries by limiting data to certain geographic regions. Therefore, it is important to make concerted efforts to advance research in these areas. There are several reasons why abstracts may not be published. Studies have reported the most common reasons for not publishing an abstract were lack of time or resources to publish, the perception results were not important enough, or the perception they would not be accepted.1, 2 However, no difference was found with regard to study quality, originality, design, sample size, and presence of a positive outcome.2 Interestin

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.