Quality and publication success of abstracts of randomized clinical trials in inflammatory bowel disease presented at Digestive Disease Week†

Abstract

BACKGROUND: The incorporation of abstracts from scientific meetings into systematic reviews and practice guidelines may reduce publication bias and delays in implementing therapeutic interventions. METHODS: All abstracts of Phase III randomized controlled trials in inflammatory bowel disease accepted at Digestive Disease Week (1998-2003) were identified. MedLine, PubMed (1997-current), EMBASE, and Google Scholar were searched for subsequent full publications. Characteristics of methodology and outcomes of the abstracts and articles were analyzed. RESULTS: The 5-year cumulative publication rate of the 82 eligible abstracts was 78%. Abstracts that presented negative results were less likely to be published than those with positive findings, particularly after the first 2 years (hazard ratio 6.45; 95% confidence interval [CI]: 2.22-18.7) with 5-year cumulative publication rates of (50% versus 91%, respectively, P < 0.001). The median time to publication was longer for negative than positive abstracts (58 versus 26 months, P < 0.001). Abstracts selected for oral presentation were more likely to be published than poster presentations (89% versus 69%; P = 0.03). A change in primary outcome results was observed in 28% (n = 18) of abstracts compared to that in the final publication, and 6% (n = 4) had a statistically significant change resulting in a change of study conclusions. CONCLUSIONS: Our findings suggest that the use of abstract data would enable detection and mitigation of publication bias. Improving the uniformity and quality of abstract reporting of randomized clinical trials at scientific meetings may facilitate their incorporation in practice guidelines and systematic reviews.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 15 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.