1. Introduction

Results of biomedical research are usually summarized in abstract form and presented at scientific meetings. Professional meetings offer a platform for exchanging scientific information from the latest research, building scientific networks, and developing new ideas for future studies. However, scientific meetings are insufficient to disseminate research findings. For example, although some scientific societies publish meeting abstracts on the Internet, these abstracts do not appear in MEDLINE or other electronic databases. Furthermore, some journals limit the citation of abstracts as references to a specified time after publication of the abstract. Conference abstracts that never achieved full publication are referred to as lost information [1]. In general, scientific information contained in meeting abstracts is assumed to be subsequently published in a peer-reviewed journal in full manuscript form within a reasonable period. Therefore, the rate of publication of meeting abstracts is considered an indicator of the scientific quality of a meeting [2, 3]. A recent meta-analysis of full publication following abstract presentations that included 425 studies found publication rates ranged from 2.6% to 90.7% (median rate 37.3%) [4].

In terms of anesthesiology as a specialty, abstract to publication rates for presentations at conferences held by associations such as the International Anesthesia Research Society, American Society of Anesthesiologists, and European Society of Anaesthesiologists have been reported to range from 0% to 53.6% [5, 6, 7, 8, 9, 10, 11, 12, 13]. Approximately 10,000 anesthesiologists from around the world attend the Japanese Society of Anesthesiologists (JSA) meeting each year and several hundred research abstracts are presented. This represents a unique opportunity to learn about novel research findings in anesthesiology. Although estimating the scientific level and output of such a large professional convention is important, there are no published data reporting the subsequent full publication rate of abstracts presented at JSA meetings. Serious concerns regarding the steep decline in anesthesia journal publications from Japan have recently been raised [14, 15, 16, 17, 18]. Therefore, we need to advance our understanding of current research output, especially from Japan. In addition, previous anesthesiology research [5, 6, 7, 8, 9, 10, 11, 12, 13] did not thoroughly examine factors predictive of subsequent publication after a meeting presentation. To promote the publication process of scientific meeting abstracts, factors associated with successful publication should be elucidated.

In this study, we sought to determine the fate of abstracts presented at JSA meetings from 2015 to 2017 by focusing on two objectives: assessing the publication rate in peer-reviewed journals and identifying factors associated with successful publication. We believe this study will provide information to facilitate the full publication process after scientific meeting presentations.

2. Methods

3. Results

3.1 Publication rate

Fig. 1 shows the study flow. Of the 3,256 JSA annual meeting abstracts from 2015 to 2017, we excluded 836 abstracts that were presented at symposia, problem-based learning discussions, refresher course seminars, sponsored seminars, and other invited lectures. The remaining 2,420 abstracts underwent PubMed® searching. We found that 489 meeting abstracts were ultimately published. Of those, we excluded two abstracts from the analysis that were retracted after publication. Therefore, the publication rate of JSA meeting abstracts was 20.1% (487 of 2,418).

Fig. 1.Study flow diagram. JSA, Japanese Society of Anesthesiologists.

3.2 Time to publication

Fig. 2 shows the estimated cumulative publication rate for JSA meeting abstracts. The median time to full publication after JSA meeting presentation was 13 months (interquartile range, 3-22 months). We also found that 81 (3.3%) meeting abstracts were published in peer-reviewed journals before the JSA annual meeting in which they were presented.

Fig. 2.Cumulative full publication of research abstracts initially presented at the Japanese Society of Anesthesiologists annual meetings 2015-2017. Event rates shown are Kaplan-Meier estimates. The inset shows the same data on enlarged y axes. Among 2,418 research abstracts initially presented at JSA annual meetings 2015-2017, 487 (20.1%) abstracts resulted in full publication within 36 months. Median time to full publication after JSA meeting presentation was 13 months (interquartile rage, 3-22 months), and 3.3% (81 of 2,418) abstracts were published before the relevant JSA annual meeting. JSA, Japanese Society of Anesthesiologists.

3.3 Journal impact factor

Fig. 3 shows the distribution of the impact factors for journals in which JSA meeting abstracts were published. The median journal impact factor was 2.07 (inter quartile range 1.45-3.34), and the range was 0-15.9. The numbers of papers published in journals with an impact factor < 1, between 1 and 3, and > 3 were 67 (13.8%), 266 (54.6%), and 154 (31.6%), respectively.

Fig. 3.Journal impact factor of published research abstracts presented at the Japanese Society of Anesthesiologists annual meetings 2015-2017. Column scatter plots represent the data distribution (circles), median (horizontal bar), and interquartile range (vertical bar). The median journal impact factor was 2.07 (inter quartile range 1.45-3.34) and ranged from 0 to 15.9. The number of papers published in journals with an impact factor < 1, between 1 and 3, and > 3, were 67 (13.8%), 266 (54.6%), and 154 (31.6%), respectively.

3.6 Subgroup analysis

As shown in Fig. 4, the cumulative publication rate significantly differed by presentation style (poster presentation, 13.8%; poster discussions, 22.8%; excellent abstracts, 47.1%; P < 0.001). Similarly, the journal impact factor for final publication was significantly higher for excellent abstracts than other two presentation styles (vs. poster presentation, P < 0.001; and vs. poster discussions, P < 0.01, Fig. 5).

Fig. 4.Cumulative full publication of research abstracts by presentation style. Event rates shown are Kaplan-Meier estimates. The P-value was calculated using means of the log-rank test based on all available follow-up data. The inset shows the same data on enlarged y axes. The number of research abstracts presented as poster presentations, poster discussions, and excellent abstracts at the Japanese Society of Anesthesiologists annual meetings 2015-2017 were 1402, 759, and 257, respectively. Of those abstracts, 193 (13.8%) poster presentations, 173 (22.8%) poster discussions, and 121 (47.1%) excellent abstracts resulted in full publication within 36 months.

Fig. 5.Journal impact factor for published research abstracts presented at the Japanese Society of Anesthesiologists annual meetings 2015-2017 by presentation style. Column scatter plots represent the data distribution (circles), median (horizontal bar), and interquartile range (vertical bar). P-values were derived from Kruskal-Wallis tests followed by Dunn’s post hoc tests with Bonferroni correction. ***P < 0.001 vs. poster presentations; ††P < 0.01 vs. poster discussions.

4. Discussion

Our analysis of more than 2,000 abstracts presented at JSA meetings (2015-2017) showed that approximately one-fifth of these abstracts were ultimately published in peer-reviewed journals within 3 years after presentation. The publication rate differed significantly according to presentation style, research type, authors’ gender, affiliation characteristics, and number of collaborating facilities.

In this study, the subsequent publication rate of JSA meeting abstracts was 20.1%. This abstract-publication rate was lower than that reported in other anesthesiology studies [5, 6, 7, 8, 9, 10, 11, 12, 13]. The reasons for those discrepancies may be multifactorial. Japan has fewer physicians per capita than most other OECD countries (https://www.oecd-ilibrary.org/sites/b33ab4c1-en/index.html?itemId=/content/component/b33ab4c1-en, accessed August 8, 2020). This means that many healthcare professionals (including anesthesiologists) in Japan are chronically exposed to a heavy workload. Most Japanese anesthesiologists also expected hold multifaceted roles encompassing clinical services, teaching residents, and ICU and operating room administration. Therefore, our study population may have difficulty finding time for research activity. This assumption was supported by the results that showed first the author’s origin (abroad) and affiliation type (medical company) were both strong predictors of subsequent publication. In addition, the current JSA board certification system warrants mention. Anesthesiology residents entering specialty training programs need to make at least one presentation at a JSA annual meeting to apply for the board certification examination. Delivering a presentation at JSA annual meeting is also important for JSA board certified anesthesiologists to obtain points for retaining their license. Some JSA meeting presentations included in this study therefore might have been intended to obtain/maintain board certification, rather than subsequent publication in international journals. These differences in meeting focus along with differences in abstract selection processes, research culture, available funds/resources, data measurement, or a combination of these factors might have resulted in a relatively lower rate of anesthesiology publications in our study compared with previous studies. However, the fact that nearly 80% of JSA meeting abstracts did not achieve full publication warrants improvement.

This study also found that the first author being male was independently associated with full publication of the abstract. Japan lags behind other countries in gender equality. According to Global Gender Gap Report 2020 by World Economy Forum (http://www3.weforum.org/docs/WEF_GGGR_2020.pdf, accessed August 8, 2020), Japan ranks 121st out of 153 countries in gender equality. In Japan, women physicians still spend more time than their male counterparts engaged in domestic work (e.g., child rearing and housekeeping) [25, 26]. Because the publication cycle requires a lot of time, women researchers in the field under study may have difficulty balancing research activity and family responsibilities. Our finding was consistent with previous studies that reported a gender gap in academic achievement [27, 28, 29, 30]. Increased institutional support and continuous, focused organizational efforts are needed to promote gender equality [27, 30].

This study also elucidated the relationship between facility characteristics and full publication. We observed that academic facilities had an improved chance of publication compared with community hospitals. In Japan, obtaining academic positions related to medicine is highly competitive [30, 31]. To obtain and remain in a tenured position in an academic facility, Japanese researchers are required to continuously publish their work in peer-reviewed journals. Our results may reflect this situation.

JSA annual meetings include three presentation styles. Usually, JSA reviewers assign the top 10% of abstracts to excellent abstracts, the following 30% of abstracts to poster discussions, and remaining abstracts to poster presentations. Our analysis also demonstrated that abstracts designated as excellent abstracts were more likely to be published than the other two presentation styles. Similarly, the journal impact factor of the final publications was significantly higher for excellent abstracts than other two presentation styles. Therefore, our results suggested that abstract reviewers discern the scientific value of abstracts well, and JSA meeting presentation designation may be used as a “test” to predict the fate of abstracts.

Consistent with prior studies [3, 6, 11], meeting abstracts reporting on basic research were more likely to be published than clinical research. In most cases, experimental studies require careful forethought and planning, funding, a fully equipped laboratory, and protected research time. We assumed that such factors collectively lead to successful publication. Multi-institutional collaboration has increased and gained importance in the domain of scientific research over the last few decades [32, 33]. This study showed that the number of collaborating facilities was associated with an increased chance of publication. Our results were consistent with a previous report by Larivière et al. [34] that found an increased number of collaborating facilities led to increased research impact.

Our study had several limitations. First, the retrospective bibliometric design meant that we could not attribute causality from the associations between any exposures and publication. Second, as with any observational study, detected associations among publication and presentation style, research type, and first author and affiliation characteristics may be confounded by other factors. Although adjustments were made using a logistic regression model, there might have been other unmeasured confounders. For example, author’s age, clinical experience, workload, job title, academic degree, marital status, and number and age of children might have affected the publication rate. Further analyses including such in-depth personal information are required to further clarify factors associated with successful publication. Finally, because we searched for published articles in a single database (MEDLINE) using a single search engine (PubMed®), some peer-reviewed manuscripts published after JSA annual meetings might have been overlooked. The publication rate detected by this study is therefore likely to be lower than the actual publication rate. However, we believe our search strategy was natural and clinically relevant because most researchers initially use MEDLINE as the most reliable source of medical knowledge. We also feel that it is unlikely that a significant difference would have been found even if other databases (e.g., Embase® and Google Scholar®) were included. For example, Berry et al. [35] found that 94% of target references were identified in their MEDLINE search for articles included in systematic review of medical imaging.

Despite these limitations, we believe our study provides an accurate depiction of the current anesthesia research activity in Japan. This study included enough abstracts to ensure model stability. Our logistic model was robust with high discrimination ability (c-statistic > 0.8), and successfully detected factors predictive of publication. Further research should seek to clarify barriers to the publication of findings presented at scientific meetings and identify objective ways to assist the publication process.

5. Conclusions

This bibliometric analysis found that approximately one-fifth of JSA meeting abstracts are subsequently published in MEDLINE-indexed journals within 3 years after the meeting. Factors independently associated with subsequent full publication are presentation style, research type (basic research), male gender, specific region, affiliation characteristics (university and medical company), and increased number of collaborating facilities. We believe this study advances our understanding of current anesthesiology research output from Japan and provides important information to guide the successful publication of research abstracts. We believe that the implications of our results would be beneficial for both our study population and other settings. Our observations should stimulate further investigations that clarify objective ways to assist the full publication process of scientific meeting abstracts.

Author contributions

YO conceived the study design, collected the data, conducted the database search and the statistical analyses, drafted the initial manuscript, and takes primary responsibility for the whole paper. MS replicated the database searching after YO and checked the data validity and reproducibility. MS also participated in drafting the paper. As a professional librarian, CS provided critical advice and made substantial contributions to the construction of the database search strategy. CS also supervised the database searching and participated in drafting the paper. KS helped in conceptualizing the study design, critically reviewed the manuscript, and participated in reviewing various drafts. SI and JK supervised the study, and participated in discussing, revising, and editing the manuscript. All authors read and approved the final version of the manuscript.

Ethics approval and consent to participate

Current Japanese ethical guidelines do not mandate ethical review for studies analyzing publicly available data. This study was therefore exempt from ethical evaluation by the Institutional Review Board at the authors’ facilities. Similarly, as this was a bibliometric analysis that did not involve human subjects, the need for informed consent was waived.

Acknowledgment

We thank Kazushige Toyama, MD, PhD, Tuneyoshi Seki, MD, PhD, Yusuke Miyazaki, MD, and Yoshihisa Fujinami, MD (from the Department of Disaster and Emergency Medicine, Kobe University, Kobe, Japan) for participating in the related discussion. We also thank Audrey Holmes, MA, from Edanz Group (https://en-author-services.edanz.com/ac) for editing a draft of this manuscript. Finally, we thank Nozomi Ono, MD (Department of Psychiatry, Hoshigaoka Hospital, Koriyama, Japan) for assisting in reviewing the manuscript.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Conflict of interest

The authors declare that there are no conflicts of interest regarding the publication of this article.

Data availability

The data used to support the findings of this study are available from the corresponding author upon request.