At the annual congress of the Swiss Societies of Cardiology and Cardiac Surgery (SSC) current research results are presented. Abstracts can be submitted to the congress website at the beginning of each year and are rated by experts in the corresponding domain. Accepted abstracts are then presented at the annual congress either in an oral session or as a poster. Acceptance of the submitted abstracts and presentation of the results of a study at the annual congress is praiseworthy, but more important is their publication in important international journals.

This study aims to give an overview of the state of scientific work in the field of cardiology in Switzerland. We determined how many of the studies presented at the four congresses were published and, if so, in which journal and with what impact factor (IF). We further looked for differences between the various cardiological domains in terms of publication rate and average IF, and whether they were accepted as a poster or as an oral presentation. Finally, we compared these data between the major Swiss hospitals.

The abstract booklets of the congresses 2011 to 2014 were downloaded as PDF files. All talks and posters were allotted to one of the following seven domains:

In most cases the classification was made according to the domains to which they were submitted. However, a few abstracts were reclassified. For example, abstracts ­concerning CABG or heart transplantation were reclassified into the category “Cardiac surgery” if they were submitted by researchers from a cardiac surgery ­department. In November 2015, all abstracts were searched for via PubMed and if they had been published, we documented the journal and the publication year. All studies that were not found in the first round were searched for again in spring 2016 and finally in February 2017; if they had been published, they were added as described above. The abstracts were initially screened for with the submitted title, then if untraceable with relevant keywords and in a third attempt with the first and/or last author name. The IF of the year in which each paper was published was recorded. As the IFs of 2016 were not released at the time of final analysis, the IF of the previous year was used, but this concerned only isolated papers. The impact factors of the journals were taken from the website http://admin-apps.webofknowledge.com. If they were not available there, they were searched for on the journal homepage. Journals for which the IF could not be determined were rated with an IF of 0.

Excluded were seven papers published more than 6 months before the corresponding SSC congress (we doubt that it was correct to submit them as abstracts to the SSC congress, according to congress rules), case reports, and all abstracts presenting national data from a multicentre study (unless the first, last or corresponding author of the published paper was from a Swiss hospital). Published and unpublished studies were then assigned to the corresponding hospital, according to the abstract booklet and the first, last or corresponding author of the paper. In some cases when more than one hospital was involved in the publication, the hospital of the last or corresponding author was chosen. For interhospital comparison we present separate data for the seven hospitals with the most successfully submitted abstracts. The remaining hospitals that submitted at least one abstract are also shown, but results have to be interpreted with caution because of mostly very small numbers of accepted abstracts. For calculation of the delay between presentation at the congress and publication, we allowed for a certain simplification. It would not be possible to determine the exact date of publication and so we set it to December of the year of publication.

We used the statistic program acula statistics calculator to determine means, medians and interquartile ranges in the different categories. A Cox regression analysis was performed to investigate whether there was a higher likelihood of publication of abstracts presented in earlier congresses. The temporal trend of the publication rates was calculated with the two-sided Cochran-Armitage Trend Test and the use of SAS Version 9.4 (SAS Institute, Cary, NC). Regardless of distribution (normal or skewed), we always present mean and median values with either the standard deviation (SD) and 95% confidence interval or the IQR (interquartile range), as to avoid confusion among the many tables and congress years that are shown.

A total of 900 abstracts were accepted in the years 2011 to 2014. Of those, 185 were excluded in accordance with the criteria described in the methods section. Of the remaining studies, 369 (51.6%) had been published as of February 2017 (see fig. 1). Most abstracts were submitted in the domains of “cardiac surgery” and “CAD” with 124 and 123 abstracts, respectively, the lowest number (76 abstracts) in “epidemiology”. The highest publication rates were seen in “heart failure” (70%) and in “biology” (65%), with the lowest in surgery (36%). The highest mean IF was recorded in “biology” at 7.8 and “heart failure” at 6.2, with the lowest in “cardiac surgery” at 2.5. The highest median IF was recorded in “biology” at 5.9 and “heart failure” at 4.2, with the lowest in “cardiac surgery” at 1.3. Looking at temporal trends, we generally noticed a slight decrease in publication rates from 54% in 2011 to 47% in 2014 and stable median IFs. In the different domains, “biology” showed a slight increase in publication rate with rather stable IFs. “Devices” exhibited a stable publication rate and increasing IFs. “Heart failure” had a stable publication rate and a decline in IFs. Of note is the exceptional high mean IF in 2012 due to one NEJM, one Circulation, and three European Heart Journal papers. “CAD” had a decrease in publication rates, but IFs ­remained stable. “Epidemiology” and “imaging” had undulating publication rates and stable IFs, and finally “cardiac surgery” an increase in publication rates and stable IFs. The domains “heart failure, “CAD”, and “imaging” showed a borderline ­decrease in numbers of accepted papers (p-values for trend 0.06, 0.08 and 0.08, respectively). More details are shown in the tables 1–7. Table 8 shows the percentage of accepted posters and oral presentations. Table 9 depicts publication rates of the different hospital groups. The mean delay between the congress and publication date was 20 months for the 2011 congress, 23 for the 2012 congress, 21 months for the 2013 congress, and 17 months for the 2014 congress. Cox regression showed no higher likelihood of publication of abstracts presented at earlier congresses (p-value 0.983).

The main finding of our analysis was the overall high quality of cardiology research in Switzerland, based on a high publication rate with good IFs of abstracts accepted for the congresses. There are, however, differences with regard to the domain of research and the performance of the individual hospitals. Results might encourage researchers in domains and/or hospitals with poorer results to further strengthen their efforts to get their work accepted in high ranking journals.

To the best of our knowledge, such a study has never been undertaken in Switzerland before; at least none has been published. The policy on accepting abstracts for national congresses in nearby countries such as Germany or France might be completely different and cannot be compared. Our results simply reflect the ­situation in Switzerland and should be seen as this.

On a European level, abstracts presented to the ESC congress 2006 have been evaluated regarding publication, IF and citation index [1]. A random selection of 10% of all abstracts submitted (n = 1002) was assessed 4.5 years after the congress. In that year, the acceptance rate was 38%, 18% accepted as oral presentations. Factors favouring acceptance were a prospective design and a higher number of patients in the study. Our study was not able to investigate these aspects, as we have no data on rejected abstracts of the congresses. Regarding full-text publication, only 38% of all accepted studies (with no difference between posters and oral presentations) were finally also published, as opposed to 52% in our study. This positive finding could be due to the fact that Swiss researches have more zeal “to take the last mile” compared with their colleagues submitting to the ESC congress, even though the ESC congress is by far larger and more renowned. At ESC level, there was only a small difference among the domains (range 24–31%) as opposed to our study (range 31–65%). Factors that favoured publication at ESC level were the academic setting of the place of research and basic research. This can also be seen in Switzerland where “biology” had the second-highest publication rate and the highest mean and median IF, and the publication rate of university hospitals exceeded that of other hospitals.

An analysis of the different years did not show a lower overall publication rate in the year 2014, which could be expected because of the shorter time frame in which the abstract could have been published. Generally, the vast majority of abstracts are published in the year of the congress, or within the following two years. As shown with the domain of “heart failure” in the year 2012, a few very high-ranked papers can heighten the mean IF of a specific year, so the median value must also be considered.

We are aware that results regarding IFs strongly depend on the IFs of the journals in a specific domain. This mainly applies to research work that is not at such an extremely high level as to be submitted to a top journal. However, overall publication rates are not affected by this limitation. Our last PubMed access was February 2017 and some papers may still be published after that date. This, however, applies to all domains. Papers in journals without an IF and/or that are not PubMed listed (e.g., this very journal) were excluded but may still have their merits. However, in the huge field of often dubious online journals it is very difficult to separate the wheat from the chaff. As we were not able to identify abstracts that were rejected, we could not analyse them regarding possible publication. Finally, the period of 4 years provides only a momentary snapshot, and results might be different in other time frames or longer observation periods.