Posters and oral presentations published as abstracts at scientific congresses are a valuable means of conveying new information rapidly and they usually represent the first step toward the dissemination of new work. Although some scientific societies publish abstracts of free papers as a supplement of a widely indexed journal, these abstracts do not appear in MEDLINE or other electronic databases. For many meetings, abstracts are published only in volumes of proceedings that are exclusively available for those who attend the meeting. Meeting abstracts are therefore of little use to the scientific community at large, and scientific data presented in abstracts remain little known unless they are eventually published as a full paper. Some organizers of meetings have begun to publish abstracts on the Internet. This may guarantee wider dissemination and a longer 'shelf life' of the abstracts. However, scientific studies often undergo changes from the time that they are first presented to final publication of the full article.
The rate of eventual publication of abstracts may be regarded as an indicator of the scientific level of a meeting and its presentations. Various authors have therefore begun to study patterns of subsequent full publication. For example, a recent meta-analysis by Scherer and Langenberg  found an average rate of full publication of 44.8% (95% confidence interval (CI) 44.0-45.6) for abstracts presented at different biomedical meetings. Another study showed the average rate of full publication of abstracts from biomedical meetings to be 41.2% . For anaesthesiology meetings, studies from Spain , Germany , and the USA [5,6] reported publication rates ranging from 17 to about 48%.
The main objective of our study was to determine the publication rate of abstracts that had been accepted at the 1995 ESA (European Society of Anaesthesiologists) meeting in Paris. The second aim was to identify features of these abstracts that may predict eventual full publication.
All abstracts of free papers (either oral presentations or posters) presented at the 1995 ESA meeting and published as a supplement to the British Journal of Anaesthesia were included. To identify all abstracts that have been subsequently published as full papers, we performed a MEDLINE (PubMed) search in December 2000, encompassing the years 1993-2000. We searched for the lead authors of the abstracts. If we failed to locate a paper, we also searched for the second, the third and the last author. If necessary, additional search strategies were used, for instance combinations of title words or keywords. One of the authors replicated the search using the biomedical database Datastar® (Service Dialog, Palo Alto, CA, USA) to check for errors and increase the reliability of the search. When a paper was retrieved, the abstract of the meeting presentation and the abstract of the published paper were compared.
Analysed descriptive variables were subsequent full publication, average time elapsed from meeting to publication, first author's country, ESA subspeciality classification (14 items) and the publishing journal. Potential predictive factors were oral or poster presentation, quality of the research design as suggested by the study architecture and ranked from lowest to highest (case report < descriptive observational study < case-control or cohort study < randomized trial), object of investigation (human, animal, equipment), presence and clarity of statistical reporting, statistically significant results, and sample size. The clarity of statistical reporting was defined as being adequate if the number of cases or groups and the statistics used were reported. A study size smaller than the median size of all studies was arbitrarily defined as small. We searched for differences between meeting abstracts and full publications, for instance authorship (number and order of authors), sample size and study groups.
Qualitative variables were expressed as absolute numbers of cases and percentages, quantitative variables as means, standard deviations (SD) and ranges. Qualitative variables were compared by calculating relative risk with 95% CI. A Wilcoxon signed rank sum test was used to compare the number of authors, and a U-test was used to analyse the quality of research design. Significance was assumed if P ≤ 0.05. Statistical packages used were SPSS 10 for Windows®, Epistat® and ICUMath® (PalmOS).
Four-hundred-and-seventy-two abstracts were published in the ESA 1995 supplement of the British Journal of Anaesthesia, and 199 (42.2%) of those were eventually published as full papers in MEDLINE-indexed journals. The average delay from meeting presentation to publication was 16.8 months (SD ± 15.6; range 24-60 months). Most abstracts (158/199, 79.4%) were published within 3 yr of the meeting (Fig. 1). Ten abstracts (5.0%) had already been published in full before the actual meeting.
Table 1 shows publication rates by country of the first author of the abstracts. Table 2 shows the number of meeting abstracts and full publication rates by subspeciality. Papers on circulation, pharmacology and intensive care issues had the highest rates. These subspecialities were also among the top five in terms of total number of congress presentations.
Full papers appeared in 63 different journals; 18 of those were anaesthesiology journals. Table 3 shows that anaesthesiology journals attracted 139 papers (69.8%). Sixty-four per cent of the papers were published in 10 European anaesthesia journals. The British Journal of Anaesthesia (29/199 papers, 14.6%) published more than any other journal. Non-anaesthesia journals that published at least two presented papers were Intensive Care Medicine, Critical Care Medicine, Chest, Journal of Applied Physiology, Transplantation Proceedings, Acta Physiologica Scandinavica, Advances in Experimental Medicine and Biology, American Journal of Respiratory and Critical Care Medicine, Infusionstherapie und Transfusionsmedizin, International Journal of Clinical Monitoring and Computing, Journal of the American College of Cardiology, Journal of Pharmacology and Experimental Therapeutics and Transplant International. Twenty of the 199 full papers (10.1%) were published in a language other than English: nine in French, five in Spanish, four in German and one each in Russian and in Hebrew.
Predictors of an abstract's subsequent full publication are shown in Table 4. Quality of research design (i.e. study architecture) was significantly associated with the publication rate. Research in animals was more likely to be published than clinical studies. Abstracts with adequate reporting of statistics were more likely to be published than those with less precise information, although the difference was not significant. No difference was found between the publication rate of reports with statistically significant and non-significant results, and between abstracts that were presented orally or as a poster. We found a tendency for abstracts reporting on studies with a small sample size to be more likely to be published than abstracts with larger sample sizes, although the difference was not significant.
The average number of authors of the abstracts was 4.5 (SD ± 1.6, range 1-9), and of full paper was 5.1 (±2.2, 1-14) (P < 0.001). In 145/199 papers (72.9%), some authors had changed between abstracts and full publication. In 43 cases (21.1%), changes concerned the first author. A subgroup of 183 papers (91.9%) adequately reported on statistical analysis. In 113 of these (61.7%), the groups and sample size in abstracts and full publications were the same. In 56 (30.6%), the sample size of the meeting abstracts was smaller and in the remaining 14 (7.7%) the sample size of the meeting abstract was larger.
The rate of eventual journal publication of research presented at the 1995 ESA meeting was slightly over 40%, a rate similar to that found by Boldt and Maleck when they studied subsequent publication of abstracts presented at both a German anaesthesiology congress and the 1994 ESA meeting . It was also similar to the reported publication rates of abstracts presented at other anaesthesia meetings [5,8] and non-anaesthesia medical specialities [1,9-13]. The pooled rate of publication derived from 46 studies by Scherer and Langenberg  was 44.8% (CI 44.0-45.6), although it is important to note these authors reported on a large range of single publication rates across studies (10-78%).
Only a few data are available that help to explain why more than half of the presented abstracts failed to be fully published. Lack of time was the most frequently reported reason by the authors of abstracts [13,14]. Most abstracts (79%) of the 1995 ESA meeting that achieved publication did so within 3 yr from the meeting. This is consistent with the reported 3 yr rate of 91% for orthopaedics , and with the 79 and 87% reported for other anaesthesia meetings . In a combined analysis of abstracts from about 50 biomedical meetings, one-third of the abstracts were published within 2 yr, one-third from the third to the fourth year, and one-third from the fifth to the eighth year . In our study, the average delay between meeting presentation and publication of an article was 16.8 months. Shorter delays have been reported for the American Society of Anesthesiologists (ASA) congress (12.5 months) , and longer delays for other meetings (22.8 months , 20.0 months ).
We found a pre-congress publication rate of 5.0%. Boldt and Maleck reported on a pre-congress publication rate of 5.7% for a German anaesthesia meeting, and of 1.8% for the 1994 ESA meeting . Hamlet and colleagues  and Gavazza and colleagues  - studying surgical meetings - reported on a 2 and 4% pre-presentation publication rate. Although all these studies report on small pre-congress publication rates, it is widely understood that presentations at a meeting do not serve a purpose once the full paper has been published; publication of an abstract may then be regarded as redundant.
The differences in publication rates related to the country of the presenting authors must be interpreted cautiously as unstudied factors may influence both the participation of an author at a meeting and subsequent publication of a presented paper. At seven ESA meetings, the countries with the highest number of participants were France, Germany, the UK, Belgium and Spain . In our study, these countries had publication rates of 54, 44, 44, 37 and 35%, respectively. However, other countries with lower participation published a higher proportion of papers. Language proficiency might affect publication rates. Fourteen of the 18 anaesthesia journals were published in English, although most presenting authors (80%) at the meeting came from countries with languages other than English. Only 20 (10.1%) of the 199 located articles were published in languages other than English. Given that authors may prefer to submit papers to their national non-English language journals that are not necessarily indexed in MEDLINE, we may have underestimated the publication rates of many countries. Using MEDLINE only to retrieve published full reports is therefore a limitation of the present study.
Our study confirms the prior finding [5,10,17] that most research that was presented at a speciality congress is subsequently published in full in a speciality journal. Full papers were published in a wide range of anaesthesia journals; however, most of the publishing journals were European. This is consistent with the fact that when abstracts are from national meetings, they are most likely to be published in full in corresponding national journals. For example, 42% of the abstracts that were published in full after their presentation at the 1994 German anaesthesia meeting were published in eight different German journals , and 59.5% of the abstracts published in full after the 1992 Spanish anaesthesia meeting were published in Revista Española de Anestesiología y Reanimación.
As in other studies, we were unable to show a difference in the rate of publication after oral or poster presentations [4,18]. However, in a systematic review, the rate of publication was higher for oral presentations . Factors that predicted subsequent publication of abstracts in previous studies were basic science [12,19] (compared with clinical investigations), sample size [20-22], use and clarity of statistics , statistically significant results , and positive results . In our study, randomized studies were more likely to be published than descriptive studies, and research based on animal subjects had a higher publication rate compared with clinical studies or studies on equipment. In contrast to other authors, we found a tendency for small sample sizes to be positively associated with publication. This result, however, was not statistically significant and therefore should be interpreted cautiously. The presence and clarity of statistics were also associated with a higher publication rate. Significant results, however, were not a predictive factor and we did not study the impact of positive results.
We found differences between meeting abstracts and abstracts of full papers, suggesting that the citation of congress presentations may be a problem. We identified differences, as did other authors [9,15], in the number of authors between the meeting abstracts and the final papers that were eventually published. In 73% the authorship had changed, and in 22% this change concerned the first author. Such changes make it difficult to retrieve research results and provide another reason why citation of meeting abstracts as references in publications should be discouraged. Differences in groups and in the number of analysed subjects were also important. Sample size changed by 37.8%, a rate very similar to the 35% reported by Murrey and colleagues . We consider it to be reasonable for authors to present preliminary study results at a meeting, and to add further cases or refine the interpretation of their results upon full publication. Others have found differences in the conclusions of abstracts and subsequent publications from 0.71%  to 4% . Given such differences, and given that most abstracts are published as full reports within 3 yr, we think that a 3 yr limit is a useful guideline for the acceptance of citing meeting abstracts, as specified by some journals in their instructions to authors. We also agree that restraint should be exercised in their citation in general.
In summary, the 42% publication rate in MEDLINE-indexed journals of research presented at the 1995 ESA meeting was similar to rates reported for other anaesthesia meetings and other specialities. Most abstracts were published in full within 3 yr after the meeting, a delay similar to other international speciality meetings.
Since 2000, the Annual Meeting Abstracts of the European Society of Anaesthesiologists have been published by the European Journal of Anaesthesiology, which is the official journal of the Society.
1. Scherer R, Langenberg P. Full publication of results initially presented in abstracts (Cochrane Methodology Review). In: The Cochrane Library,
Issue 2, 2002. Oxford, UK: Update Software.
2. Von Elm BE, Poglia G, Walder B, Tramèr MR. Fate of abstracts submitted to biomedical meetings: a systematic review. Fourth International Congress on peer review. Barcelona, September 2001: http://www.ama-assn.org/public/peer/peerreviewcongress.pdf/
3. Castillo J, Garcia-Guasch R, Cifuentes I. Publicaciones derivadas de las comunicaciones libres del XX Congreso de la Sociedad Española de Anestesiología y Reanimación (Anestesia '92). Rev Esp Anestesiol Reanim
4. Boldt J, Maleck W. Schicksal wissenschaftlicher Beiträge auf grossen Anästhesiekongressen. Anaesthesist
5. Meranze J, Ellison N, Greenhow DE. Publications resulting from anesthesia meeting abstracts. Anesth Analg
6. Sebel PS, Duensing SJ. Abstracts from the IARS clinical and scientific congress: proportion published and time to publication. Anesth Analg
7. European Society of Anaesthesiologists. Br J Anaesth
(Suppl 1): A1-A474.
8. Yentis SM, Campbell EA, Lerman J. Publication of abstracts presented at anaesthesia meetings. Can J Anaesth
9. Liu L, Danziger RS. Fate of conference abstracts. Nature
10. Hamlet BWP, Fletcher A, Meals RA. Publication patterns of papers presented at the Annual Meeting of the American Academy of Orthopedic Surgeons. J Bone Joint Surg Am
11. Goldman L, Loscalzo A. Fate of cardiology research originally published in abstract form. N Engl J Med
12. Gavazza JB, Foulkes GD, Meals RA. Publication pattern of papers presented at the American Society for Surgery of the Hand Annual Meeting. J Hand Surg
13. Weber EJ, Callaham ML, Wears RL, Barton C, Young G. Unpublished research from a medical specialty meeting. JAMA
14. De Bellefeuille C, Morrison CA, Tannock IF. The fate of abstracts submitted to a cancer meeting: factors which influence presentation and subsequent publication. Ann Oncol
15. Murrey DB, Wright RW, Seiler JG, Day TE, Schwartz HS. Publication rates of abstracts presented at the 1993 Annual Academy Meeting. Clin Orthop
16. Canet J, Bassons J. What countries are contributing scientifically to the ESA meetings? Eur J Anaesthesiol
(Suppl 21): A13, 4.
17. Wang JC, Yoo S, Delamarter RB. The publication rates of presentations at Major Spine Specialty Society Meetings (NASS, SRS, ISSLS). Spine
18. Loevy HT, Curtis KL, Goldberg AF. Conversion rate of research abstracts to publications in pediatric dentistry. Pediatr Dent
19. Elder NC, Blake RL. Publication patterns of presentations at the Society of Teachers of Family Medicine and North American Primary Care Research Group Annual meetings. Fam Med
20. Landry VL. The publication outcome for the papers presented at the 1990 ABA conference. J Burn Care Rehabil
21. Callaham ML, Wears RL, Weber EJ, Barton C, Young G. Positive-outcome bias and other limitations in the outcome of research abstracts submitted to a scientific meeting. JAMA
22. Chalmers I, Adams M, Dickersin K, et al.
A cohort study of summary reports of controlled trials. JAMA