Transition From Meeting Abstract to Full-length Journal Article for Randomized Controlled Trials

CONTEXT: Not all research presented at scientific meetings is subsequently published and, even when it is, there may be inconsistencies between these results and what is ultimately printed. Although late-breaking trials sessions are now integrated into several major scientific meetings and the results are often promptly and prominently communicated, no studies have examined the publication fate and degree of consistency between meeting abstracts or presentations and subsequent full-length article publications for randomized controlled trials (RCTs) presented at these sessions. OBJECTIVE: To compare RCT abstracts presented in the late-breaking trials session vs other sessions at a major scientific meeting and subsequent full-length publications. DESIGN: RCTs were identified by hand searching abstract proceedings booklets and related Web sites for the American College of Cardiology scientific meetings (1999-2002). Subsequent full-length articles were identified via electronic databases. MAIN OUTCOME MEASURES: Publication fate and degree of consistency between meeting abstract results and subsequent full-length publication results. RESULTS: The 86 late-breaking RCTs were significantly larger (median, 2737 patients vs 896; P<.001), were more likely to be preceded by a published design paper (27 31% vs 13 13%; P = .002), had higher quality scores when eventually published (mean Jadad score 2.69 vs 2.19; P = .01), and were less likely to report favorable results for the intervention than the 100 randomly chosen comparison RCTs presented in other sessions (50 58% vs 75 75%; P = .01; odds ratio 0.46; 95% confidence interval, 0.24-0.90). RCTs presented at the late-breaking trials sessions were significantly more likely to be published (79 92% vs 69 69%; P<.001) and appeared earlier after presentation (median 11.5 months vs 22.0 months; P<.001) than RCTs presented in other sessions, an association that persisted even after adjusting for sample size, conclusion of study, and RCT design: adjusted hazard ratio, 1.80 (95% confidence interval, 1.24-2.61). Sixty (41%) of the 148 RCTs that were subsequently published exhibited discrepancies between the efficacy estimate reported in the meeting abstract vs the one reported in the full-length article for the primary outcome. The mean change in effect was 0.44 SDs and in 20 cases (14%), the point estimate was statistically significant in only 1 member of the pair. The discrepancy rate was the same for late-breaking RCTs as for RCTs presented in other American College of Cardiology sessions (P = .92). CONCLUSIONS: Late-breaking trials were larger, more likely to be preceded by a design paper, and less likely to report positive results than RCTs presented at other sessions, but discrepancies between the meeting abstract results and subsequent full-length publication results were common even for late-breaking trials.