From conference abstract to full publication: what are the facts, figures and implications of failure to publish in full?
Posted on 11th February 2019 by Neelam Khan
In this blog, Neelam Khan, a fourth year dental student at the University of Manchester, looks at the latest Cochrane evidence on the full publication of study results initially presented in abstracts and reflects on the implications of the findings.
What is an abstract?
An abstract is a concise and clearly structured summary of the main features and results of a study. It serves two key purposes:
- It allows the reader to make a quick decision on whether a paper is relevant to their interests and worth reading in full.
- Online databases that store innumerable research papers often use abstracts to organise journal articles under key subject headings in a process known as indexing. This makes studies easier to find when searching databases, for example, to conduct a systematic review.
When investigators carry out research, they will write and submit an abstract for their study to be presented at conferences. However, not all studies manage to get published in full as journal articles. This means that their abstracts are usually just collated as conference proceedings, which are not easy to find.
What are the facts and figures?
A Cochrane review examining the proportion of conference/scientific meeting abstracts that are subsequently published in their entirety has recently been updated. Having reviewed 425 reports, which looked at a staggering total of 307,028 abstracts across several medical subspecialties, the authors found that less than half of all studies (46.4%) make it to full publication within 10 years of being presented at a conference. The consequence of this is that a considerable amount of research effectively goes to waste since the information produced in the vast majority of studies is poorly disseminated. What’s more, these studies risk being missed out of systematic reviews that rely only on fully published data.
Disappointingly, this issue does not appear to be on the decline either! This is evident when the results from this review update are compared with the previous 2007 version, which found that just over half (52.6%) of the studies whose abstracts are in conference proceedings are fully published within 9 years.
Interestingly, the review also explored factors that are linked with increased likelihood of full publication and over a dozen factors* were identified including the type of research, the report authors, and the perceived quality of the study. Of concern is that two important factors that were associated with a higher probability of publication were the presence of ‘positive’ results in the abstract and if the report originated from a native English-speaking country or was written in English.
Why is this a problem, you ask?
Let’s tackle the issue of positive results first. In effect, there is a bias in published literature that favours studies with a positive outcome. Studies that show no evidence of an effect or ‘negative’ results are not fairly represented in published data. The consequence of this is that the information available, for instance regarding an intervention for a particular disease, is inaccurate and misleading. This has a knock-on effect on the findings of systematic reviews that synthesise only published data.
As for the issue of reports written in English or from an English-speaking country, this is known as language bias. Although it is not possible to ascertain the exact reason why this bias exists in publishing, the Cochrane review authors discuss a multitude of possible explanations. Nevertheless, the result of the bias is the same; that is, systematic reviewers who seek to answer a research question pertinent to a population or setting in a non-native English-speaking country may be missing out relevant studies, owing to the failure to fully publish the report.
So, what are the implications of these findings?
Essentially, it all boils down to the potential for systematic reviews of the published literature having incomplete information or an unbalanced view of the subject in question. In healthcare-related reviews, this means that the estimates of effects for certain interventions may be inaccurate or biased. This, in turn, means that decisions based on these findings may not be as beneficial as anticipated.
Since clinicians and guideline developers use systematic reviews to make evidence-based decisions, it is important that the published literature is free from any plausible bias. It also highlights the importance for reviewers to search for unpublished and ongoing trial data too to contribute to their synthesis of the evidence.
* A list of the key factors associated with full publication, identified in the Cochrane review, are detailed below:
- Studies with positive results
- Studies with larger sample sizes
- Studies with abstracts presented orally are more likely to be published than those presented as posters
- Studies accepted for presentation at a meeting are more likely to be published than those not accepted
- Studies describing basic science are more likely to be published than those describing clinical research
- Studies describing randomized trials are more likely to be published than those describing other types of studies
- Studies that took place in multiple centres are more likely to be published than those at a single centre
- Studies classified as ‘high quality’ are more likely to be published than ‘low quality’ studies
- Studies with authors from an academic setting are more likely to be published than those with authors from other settings
- Studies considered by the report authors to have a high impact are more likely to be published than other studies
- Studies with funding source reported are more likely to be published than those not reporting funding
- Studies originating in North America or Europe are more likely to be published than those originating elsewhere
- Studies from English‐speaking countries are more likely to be published than studies originating elsewhere