Comparison of conference abstracts and full-text articles of randomized controlled trials in the field of pain: reporting quality and agreement in results

Tags: Oral
Dragicevic K1, Jelicic Kadic A1, Saldanha I2, Puljak L1
1Cochrane Croatia, Croatia, 2Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland, USA

Background: According to current standards, systematic reviews should search for unpublished studies, i.e. grey literature. There is debate, however, about whether studies available only as conference abstracts ('abstracts') should be included at all in systematic reviews because it may be difficult to assess risk of bias and extract data accurately from the limited information available in abstracts. Additionally, discrepancies between conference abstracts and full publications of abstracts of the same randomized controlled trials (RCT) have been documented in various research fields.

Objectives: 1) to quantify agreement between results of primary outcomes of RCTs reported in abstracts presented at the four most recent World Congresses on Pain (WCP) and their corresponding full publications; and 2) to use the CONSORT (Consolidated Standards of Reporting Trials) for Abstracts checklist to examine the completeness of reporting in those abstracts.

Methods: Single screening with verification was conducted for all abstracts to determine which abstracts describe RCTs. Two independent authors identified corresponding full-text reports through October 2015 by electronic searches in PubMed, Google Scholar, and Embase, as well as by emailing authors. Data about the primary outcomes will be extracted from each abstract and full publication, including the outcome domains measured and numerical results reported. We will categorize any discordance (disagreement) between the primary outcome's results in the abstract and its corresponding publication as qualitative (difference in direction of effect estimate) or quantitative (no difference in direction of effect estimate). Two authors independently will evaluate all abstracts against all 17 recommended checklist items in CONSORT for Abstracts. All discrepancies will be resolved by consensus or, if necessary, discussion with a third author.

Results and conclusions: As far as we know, this is the first analysis examining agreement in conference abstracts and full publications describing RCTs addressing pain. We will present our detailed results at the Colloquium.