An interrupted time series analysis showed suboptimal improvement in reporting quality of trial abstract

Abstract

Objective: To assess and compare the immediate and long-term change in reporting quality of randomized controlled trial (RCT) abstracts published in Pediatrics, The Journal of Pediatrics, and JAMA Pediatrics before and after the publication of Consolidated Standards of Reporting Trial (CONSORT)-abstract statement.

Study design and settings: Study had "Interrupted time-series" design. Eligible RCT abstracts were retrieved by PubMed search in two study periods from January 2003 to December 2007 (pre-CONSORT) and January 2010 to December 2014 (post-CONSORT). These abstracts were matched with the CONSORT checklist for abstracts. The primary outcome measure was CONSORT-abstract score defined as number of CONSORT items correctly reported divided by 18 and expressed as percentage. The mean percentage scores were used to compare reporting quality between pre- and post-CONSORT using segmented linear regression.

Results: A total of 424 RCT abstracts in pre-CONSORT and 467 in post-CONSORT were analyzed. A significant change in slope of regression line between two time periods (0.151 [confidence interval CI, 0.004-0.298], P = 0.044) was observed. Intercepts did not show a significant difference (-2.39 [CI, 4.93-0.157], P = 0.065).

Conclusion: The overall reporting quality of RCT abstracts in the high-impact pediatrics journals was suboptimal (<50%); however, it improved when assessed over a 5-year period, implying slow but gradual adoption of guideline.

Keywords: CONSORT extension for abstracts; CONSORT statement; Interrupted time series; Quasi-experimental design; Randomized controlled trials; Segmented linear regression.