Canadian Journal of Cardiology 29 (2013) 1520e1523
Training/Practice Training/Health Policy and Promotion
Rate and Predictors of the Conversion of Abstracts Presented at the Canadian Cardiovascular Congress Scientific Meetings to Full Peer-Reviewed Publications Wael Abuzeid, MD, FRCPC,a Emil L. Fosbøl, MD, PhD,b Philip L. Fosbøl, PhD,c Marie Fosbøl, MD,d Sanaz Zarinehbaf, MD, CCFP,e Heather Ross, MD, MSc, FRCPC,f Dennis T. Ko, MD, MSc, FRCPC,a,g,h,i Maria C. Bennell, MSc, MPH,g and Harindra C. Wijeysundera, MD, PhD, FRCPCa,g,h,i a
Schulich Heart Center, Sunnybrook Health Sciences Center, Division of Cardiology, Department of Medicine, University of Toronto, Toronto, Ontario, Canada b
c
University Hospital of Copenhagen, Rigshospitalet, University of Copenhagen, Copenhagen, Denmark
Department of Chemical and Biochemical Engineering, Center for Energy Resources Engineering (CERE), Technical University of Denmark (DTU), Kongens Lyngby, Denmark d
Department of Clinical Physiology and Nuclear Medicine, University Hospital Hvidovre, Hvidovre, Denmark e
f
Department of Family Medicine, University of Toronto, Toronto, Ontario, Canada
University Health Network, Division of Cardiology, Department of Medicine, University of Toronto, Toronto, Ontario, Canada g
Sunnybrook Research Institute, University of Toronto, Toronto, Ontario, Canada h
i
Institute for Clinical Evaluative Sciences (ICES), Toronto, Ontario, Canada
Institute for Health Policy, Management and Evaluation (IHPME), University of Toronto, Toronto, Ontario, Canada
ABSTRACT
RESUM E
The rate of conversion of abstracts presented at scientific meetings into peer-reviewed published manuscripts is an important metric for medical societies, because it facilitates translation of scientific knowledge into practice. We determined the rate and predictors of conversion of scientific abstracts presented at the Canadian Cardiovascular Congress (CCC) from 2006 to 2010 into peer-reviewed article publications within 2 years of their initial presentation. Using a previously validated computer algorithm, we searched the International Statistical Institute Web of Science to identify peer-reviewed full
sume s pre sente s aux re unions scientiLe taux de conversion des re value s par des pairs est un fiques en publication de manuscrits e te s me dicales, parce qu’il facilite la important paramètre des socie transmission des connaissances scientifiques à la pratique. Nous termine le taux et les pre dicteurs de conversion des re sume s avons de sente s au Congrès canadien sur la sante carscientifiques pre diovasculaire (CCSC) de 2006 à 2010 en publications d’articles value s par des pairs dans les 2 ans suivant leur pre sentation initiale. e jà valide , nous avons En utilisant un algorithme informatique de
Medical conferences act as a venue for dissemination of knowledge and latest research among peers.1 However, many abstracts presented at conferences are never published as peer-
reviewed articles and might communicate premature and sometimes inaccurate results.2 It has been suggested that a measure of the robustness of research presented at scientific meetings is the successful publication in a scientific journal after rigourous peer review.3,4 The Canadian Cardiovascular Congress (CCC) is the major annual scientific meeting of the cardiovascular community in Canada. As such, it represents the primary national venue for abstract presentation for Canadian researchers. To the best of our knowledge, there are no data on the rate of conversion of abstracts accepted for presentation at the CCC into peer-reviewed published articles.
Received for publication May 28, 2013. Accepted June 24, 2013. Corresponding author: Dr Harindra C. Wijeysundera, Schulich Heart Center, Sunnybrook Health Sciences Centre, 2075 Bayview Ave, Suite A202, Toronto, Ontario M4N 3M5, Canada. Tel.: þ1-416-480-4527; fax: þ1-416-480-4657. E-mail:
[email protected] See page 1523 for disclosure information.
0828-282X/$ - see front matter Ó 2013 Canadian Cardiovascular Society. Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.cjca.2013.06.012
Abuzeid et al. Conversion of CCC Abstracts Into Manuscripts
1521
manuscript publications of these abstracts. A multivariable logistic regression was used to identify independent factors associated with successful publication. From 2006 to 2010, 3565 abstracts were presented at the CCC. Overall 24.1% of presented abstracts were published within 2 years of the conference. Mean impact factor for publications was 5.2 (range, 0.4-53.2). The type of presentation (for poster vs oral; odds ratio, 0.71; 95% confidence interval, 0.60-0.83; P < 0.001) and category of presentation (P < 0.001) were significantly associated with successful publication. Late breaking abstracts and those related to cancer and clinical sciences were more likely to be published, compared with prevention, vascular biology, and pediatrics. In conclusion, the publication rate at the CCC is only marginally lower than that reported for large international North American and European cardiology conferences (30.6%). Efforts should focus on several identified barriers to improve conversion of abstracts to full report publication.
sur le Web of Science de l’Institut international de statistique cherche grales de manuscrits e value s par afin de relever les publications inte sume s. La re gression logistique multivarie e a e te des pairs de ces re e pour de terminer les facteurs inde pendants associe s à une utilise ussie. De 2006 à 2010, 3565 re sume s ont e te pre sente s publication re sume s pre sente s ont e te au CCSC. Dans l’ensemble, 24,1 % des re s dans les 2 ans suivant le congrès. Le facteur d’impact moyen publie tendue, 0,4-53,2). Le type de te de 5,2 (e des publications a e sentations (par affiches vs orales; ratio d’incidence approche , 0,71; pre gories intervalle de confiance à 95 %, 0,60-0,83; P < 0,001) et de cate sentation (P < 0,001) a e te significativement associe à une de pre ussie. Les re sume s de dernière minute et ceux lie s au publication re te plus susceptibles d’être cancer et aux sciences cliniques ont e s comparativement à ceux lie s à la pre vention, à la biologie publie diatrie. En conclusion, le taux de publication au vasculaire et à la pe gèrement infe rieur à celui rapporte par les grands CCSC est à peine le rique du Nord et de congrès internationaux de cardiologie de l’Ame s sur les nombreux l’Europe (30,6 %). Les efforts devraient être axe termine s afin d’ame liorer les taux de conversion des obstacles de sume s en publication inte grale des rapports. re
Accordingly, to address this gap in knowledge, we evaluated factors affecting the publication of abstracts presented at the CCC meetings. Our goal was to compare the publication rate for CCC abstracts to international cardiology conferences and identify potential barriers to publication.
extended for 2 calendar years. The CCC is held in October; therefore, the search period spanned a 28-month period from September of the index year (September 2006), to December 31 two years later (December 2008). This time horizon is similar to previous studies.4 The details of this algorithm are described elsewhere.4 Briefly, the algorithm had 3 iterations. In the first iteration, the last name of the first and last author and 3 keywords from the abstract title were used in the search.4 A second and third iteration were performed if more than 1 publication was retrieved.4 In these, either additional keywords or additional authors were used to refine the search. In their previous publication, Fosbol and colleagues validated this search using a manual search of 200 abstracts, and found sensitivity of 92% (confidence interval [CI], 0.83-0.96), a specificity of 90% (CI, 0.82-0.95), a negative predictive value of 94% (CI, 0.86-0.97), and a positive predictive value of 88% (CI, 0.79-0.94).4 We did not repeat this validation exercise in our study.
Methods Abstract extraction We identified all abstracts presented at the CCC between 2006 and 2010, using the annual CCC abstract supplement of the Canadian Journal of Cardiology. These included abstracts submitted as part of the Canadian Cardiovascular Society, the Canadian Society of Atherosclerosis, Thrombosis, and Vascular Biology, the Canadian Society of Cardiac Surgeons, and the Canadian Pediatric Cardiology Association. This supplement also contains abstracts presented at the Canadian Council of Cardiovascular Nurses, which were not included in our analysis because they had a separate peer evaluation process. Two authors (W.A. and S.Z.) used a standardized data entry form to extract information about abstract title, author names, institutional location, and type of presentation (oral vs poster). Abstracts were categorized into 1 of 18 scientific categories based on the original scientific program. These categories included coronary artery disease, arrhythmia, clinical sciences, basic sciences, and late-breaking trials (see Table 1 for full list). In addition, we contacted the CCC secretariat to obtain data on the rate of abstract submissions to acceptance for each year of interest, and the mean scores (out of 10) for accepted abstracts. Search algorithm Using an automated computer algorithm, the International Statistical Institute Web of Science search engine (Thomson Reuters)5 was used to search for publication outcomes of the abstracts presented at CCC. In addition, we obtained the journal impact factor of published abstracts. The search period began 1 month before to the index abstract presentation, and
Statistical analysis Appropriate univariate tests were used to compare continuous (t test and analysis of variance) and categorical variables (c2). A multivariable logistic regression model was used to identify factors independently associated with publication. The variables used were year of publication, category of presentation, type of presentation (oral vs poster), and institutional location. Statistical analyses were conducted using SAS 9.3 (SAS Institute, Cary, NC). All P values were 2-sided with a value of < 0.05 considered statistically significant. Results At the 5 annual CCC scientific conferences between 2006 and 2010, 3565 abstracts were presented, from a total of 4465 abstract submissions (79.8% acceptance rate). Of the presented abstracts, 62.9% were poster presentations (n ¼ 2243) and 37.1% were oral presentations (n ¼ 1322). The most frequent categories of abstracts were coronary artery disease (20.8%), followed by surgery (14.1%), hypertension (12.0%),
1522
Canadian Journal of Cardiology Volume 29 2013
Table 1. Abstract characteristics Study Characteristics Total abstracts presented Abstract presentation by year 2006 2007 2008 2009 2010 Type of abstract presentation Poster Oral Category of presentation Arrhythmia Basic Science CAD Cancer CVD Cardiac Rehabiliation Clinical science Congenital Genetics HF Hypertension Imaging Interventional cardiology Late-breaking Pediatrics Prevention Surgery Vascular biology Abstract presentation by country Canada United States Other countries Abstract presentation by province Alberta British Columbia Manitoba New Brunswick Newfoundland Nova Scotia Ontario Quebec Saskatchewan Yukon Territory
n
%
3565 569 758 862 741 635
16.0 21.3 24.2 20.8 17.8
2243 1322
62.9 37.1
377 154 741 6 15 17 25 222 29 333 426 254 227 29 5 191 502 12
10.6 4.3 20.8 0.2 0.4 0.5 0.7 6.2 0.8 9.3 12 7.1 6.4 0.8 0.1 5.4 14.1 0.3
3390 91 84
95.1 2.6 2.4
446 247 101 5 15 124 1411 995 44 2
13.2 7.3 3.0 0.2 0.4 3.7 41.6 29.4 1.3 0.1
CAD, coronary artery disease; CVD, cardiovascular disease; HF, heart failure.
and arrhythmia (10.6%). Most abstracts were from Canada (95.1%), of which 41.6% were from Ontario (Table 1). Overall 24.1% of presented abstracts at the CCC between 2006 and 2010 were published within 2 calendar years after presentation at the conference. The mean journal impact factor of the published abstracts was 5.2 (range, 0.4-53.3), with a median value of 3.7 (Table 2). Of the abstracts that were published in the study time frame, the highest proportions were published in Circulation (4.7% of publications; n ¼ 40) and the Canadian Journal of Cardiology (4.2% of publications; n ¼ 37). Substantial variation was seen in the proportion of abstracts within a scientific category that were ultimately published (Supplemental Fig. S1); ranging from no publications in the pediatrics category to 66.7% in the cancer category. In our multivariable logistic model, scientific category (P < 0.001) and the type of presentation (adjusted odds ratio for poster vs oral: 0.71, 95% CI, 0.60-0.83; P < 0.001) were
independent predictors of successful publication outcome. In contrast, the year of the conference (P ¼ 0.48) and institutional location (P ¼ 0.99) were not statistically significant. Discussion In this study, we found that approximately 1 in 4 abstracts presented at the CCC were published as peer-reviewed full article publications within 2 years. The type of presentation and category of presentation were significantly associated with improved rates of final publication. Fosbol et al. reported that 30.6% of presented abstracts at European and American cardiology meetings were published within 2 years of the conference, ranging from 34.5% for the American Heart Association (AHA) to 29.5% for the American College of Cardiology (ACC) to 27% for the European Society of Cardiology (ESC).4 CCC abstracts only fall marginally behind these large international meetings in being published. This is despite the CCC being a substantially smaller, national meeting. These conferences explicitly prohibit abstracts that have been previously presented; as such, Canadian investigators might preferentially present higher quality work at the larger international meetings. This is suggested by the fact that median impact factor of subsequent CCC publications (3.7) was lower than that reported for the AHA (4.8), ACC (4.0), and ESC (3.9).4 Considering these facts, it should be reassuring to CCC organizers that the conversion rate of CCC abstracts is at its current level. Nonetheless, our results suggest that there is room for improvement. All abstracts at scientific meetings undergo peer assessment. The ratio of submitted to accepted abstracts is a measure of the rigour of that initial peer assessment. We found that almost 80% of submitted abstracts were ultimately accepted for presentation. This is in contrast to the 30% of abstracts submitted to AHA, ACC, and ESC that were accepted.4 This might reflect that a much higher volume of abstracts are submitted to the larger international meetings, permitting a more restrictive evaluation process, resulting in better quality abstracts that have a greater likelihood of subsequent publication. Indeed, the fact that poster presentations at the CCC had lower publication rates might reflect their lower quality. The mean score for oral abstracts were 7.85 out of 10, compared with 7.05 for poster abstracts. The range in publication rates between scientific categories is
Table 2. Conversion of presented abstracts Outcome Total number of publications Publications per year of presentation* 2006 2007 2008 2009 2010 Impact factor of published presentations Mean Median Range
n (%) 858 (24.1) 135 178 190 197 158
(23.7) (23.5) (22.0) (26.6) (24.9)
5.2 3.7 0.4-53.3
* Proportion based on total number of abstract presentations per year.
Abuzeid et al. Conversion of CCC Abstracts Into Manuscripts
noteworthy and suggests that attention should be given to some areas to improve publication. Our study should be interpreted in light of several limitations that merit discussion. First, we were only able to extract limited information on the actual abstracts. For example, we were unable to determine whether the abstracts showed negative or positive results. We were not able to determine if an abstract was designated as featured research or being considered for a trainee award, both of which might have increased the likelihood of publication. Furthermore, we were unable to record the level of training of the presenter (clinical trainee, MSc, or PhD) or sex information. Second, the International Statistical Institute system provides the date of publication as the date for when the article was in print. In journals with early online publication, we might have overestimated the time from presentation to publication. Finally, we only examined the rate of conversion of abstracts into full peer-reviewed articles. We recognize there are other ways of disseminating research other than peer-reviewed articles, such as Web pages, and reports. Grey literature of this nature was not the focus of our study, but nonetheless might represent effective knowledge translation. In conclusion, this study found that approximately 1 in 4 abstracts presented at the CCC were converted into a full peer-reviewed article. We believe our work identifies potential areas for CCC organizers to concentrate efforts to improve final publication of abstracts and thus improve the dissemination of research.
1523
Disclosures The authors have no conflicts of interest to disclose References 1. Green M, Owen D. Are international medical conferences an outdated luxury the planet can’t afford? BMJ 2008;336:1466-7. 2. Ioannidis JP. Are medical conferences useful? And for whom? JAMA 2012;307:1257-8. 3. McCormick MC, Holmes JH. Publication of research presented at the pediatric meetings. Change in selection. Am J Dis Child 1985;139:122-6. 4. Fosbol E, Fosbol P, Harrington R, Eapen Z, Peterson E. The conversion of cardiovascular conference abstracts to publications. Circulation 2012;126:2819-25. 5. ISI Web of Knowledge. Available at: webofknowledge.com. Accessed April 1, 2013.
Supplementary Material To access the supplementary material accompanying this article, visit the online version of the Canadian Journal of Cardiology at www.onlinecjc.ca and at and at http://dx.doi. org/10.1016/j.cjca.2013.06.012.